Home

Angel Ramos de Miguel 29 July 2013 - Acceda

Contents

1. Save the receive waypoint Set actuators position Update the emition time Calculate the new bearing to the waypoint Fuzzy Logic to set actuators Set actuator position Update and send sensors data Figure 6 4 On Board Flowchart 74 CHAPTER 6 CONTROL SYSTEM 6 2 2 Initialization The initialization of the system is carried out normally with the vessel at shore but could be done remotely as far as the radio link is possible In this phase the operational state of all onboard subsystems are verified and some on board sensors may be calibrated namely the wind vane and the inclinometers The calibration steps can be omitted using the base station interface First radio communication SD logging and battery levels are checked and afterwards on board power regulators are switched on Then the GPS receiver is configured and the elevation and signal strength masks are con figured in order to minimize noise in GPS readings Once the GPS is config ured a first valid fix is awaited and then it will wait 30 additional seconds to stabilize the GPS measurements Finally the fuzzy logic control system is initialized An optional final stage in initialization deals with the calibration of some sensors offsets It requires to keep the sailboat in a horizontal position with 0 of pitch and roll and the wind vane pointing forward This is done only once at the beginning of the experiment but can be avoided if previously
2. 4 GPS Measurements The GPS available for the Waspmote is a slow sensor and it can be quite inaccurate if it is not properly configured We performed some previous tests to characterize the level of noise in GPS position estimates placing a static GPS receiver in an open area witha clear and equilibrated view of the sky The results of these tests determined that better results were obtained if satellites that were too low over the horizon or provided very weak signal were not used to compute the position fix as recommended in the GPS literature Figure 7 5 depicts a typical downwind upwind test course under constant bearing where the elevation mask was set to 15 and the signal strength mask was set to 20 dBi the standard GPS configuration in the A TIRMA This figure shows a vast majority of clean position estimates that have been characteristic of all sea trials under this configuration Nevertheless we developed a Kalman filter to combine the measurements Latitude 28 1323 28 1323 28 1323 28 1322 28 1322 28 1322 28 1322 28 1322 28 13 28 13 28 13 21 21 4 284 100 CHAPTER 7 RESULTS of the GPS with the bearing angle provided by the compass and vessel s estimated speed to achieve more robust and faster position estimates Even though the Kalman filter could provide some immunity against occasional large GPS errors finally this filter has not been integrated in the current version of the on boar
3. 7 3 NAVIGATION TASKS 105 Latitude OverHeeling Position 28 1322 Sail Position 28 1322 28 1322 28 1322 28 1322 28 1321 28 1321 15 4283 15 4283 15 4283 15 4282 15 4282 15 4282 15 4282 15 4282 15 4282 15 4282 Longitude Figure 7 9 Over Heeling Position Effect OverHeeling 400 350 300 Compass degrees Poll Degrees a m nm o an o an o o o o mn o 0 20 40 60 80 100 120 140 160 180 200 Cycles Figure 7 10 Over Heeling Effect 106 CHAPTER 7 RESULTS 7 3 4 Station Keeping Task Station keeping is the capacity of any vehicle to maintain is position close to a designated location It has also been proposed as a test task for WRSC 13 In that case it is required to go autonomously inside a square of side of 50 m and stay in the middle for a certain amount of time To test the station keeping task we set two waypoints with a distance of the center of the square of 1 25 meters and we kept that point during 5 minutes Figure 7 11 and Table 7 5 summarize the results Table 7 5 Mesurements Distance between waypoints 2 50 m Minimum distance to the center 0 38 m Maximum distance to the center 7 83 m Mean distance to the center 4 00 m Standard deviation 1 79 m Minimum square 10 45 x 13 33 m We tried also this task using only one waypoint and kept the task active during 12 minutes with the results depicted Figure 7 12 and Table 7 6 Table 7 6 Mesurements Minimu
4. Sak ol Base Station GUls ars Era sh yey uae eon Beka ees ee SS 54 MBeesgateway Li A A a Soa oS 55 Logitech WingMan Cordless 2 2 0 ee eee ee 56 On Board Flowchart so e e ios orea tiano o e ae e e a a 73 Fuzzy input Desired heading s e u e a a eee ee ee 75 Fuzzy Input Current rudder angle ee 76 Fuzzy output Commanded rudder angle 76 Fuzzy input Point of sailing s ooo 78 Fuzzy input Rollangle a o gor eoo sou aai ai e i 00000007 79 Fuzzy output Sail position s a so e e a 79 Control architecture ee 81 Apparent wind error 1 soosoo e e e e 84 Timing with and without code optimizations 92 Timing with and without code optimizations 94 Wind load estimation os scs s non pae ne io nana a a 96 Wind Eter segne o ak AA a aar E RR 98 GPS Accuracy o ona A A N ha a 100 WRSCrar na A a D i a Ge ee a ee Be 101 An acCouraty t st i e a a i A ee BO es Be E Sa 102 Return To Homes fs ea ati pt a ek aod a OS 103 Over Heeling Position Effect e ee eee 105 Over Heeling Effect cios hi ako a oe ee ee a 105 Station keeping results ee ee 107 Second station keeping results 2 107 Servo Battery nue pate a a A a g 108 Waspmote Battery osoo a a 109 List of Tables 1 1 3 1 3 2 3 3 3 4 3 9 3 6 4 1 4 2 5 1 5 2 6 1 6 2 6 3 6 4 6 5 6 6 6 7 6 8
5. class sailboat severely restricts the volume and weight of sensors and control elec tronics that can be installed on board These restrictions have a negative impact in terms of computing power and sensors that can be used in the development of the sailboat s control system The main limitations of the control system described in this thesis are its scarce computing power and reduced RAM memory These restrictions 111 112 CHAPTER 8 CONCLUSIONS have demanded a careful analysis and design of the control system to make it fit within the microcontroller memory and processing power trying at the same time to reduce the span of a control cycle as much as possible This control system is very similar in scope to that described in 48 where it was presented a control system for autonomous sailboats based on a 50 MHz 64KB RAM Cortex M3 ARM7 processor board The main difference between both systems aside from the smaller computing power and memory of our system is that our sailing control system is completely embedded on the on board processor while in 48 the sailboat controller executes in a laptop outside of the boat Perhaps the most fragile element of the whole system is the wind vane as noted by many others 49 Wind vanes with movable mechanical parts are intrinsically prone to failure Whilst commercial solutions exist for full scale sailboats they are unpractical for a boat of small dimensions Some solutions have been e
6. com p arduino issues detail id 857 Contents Abstract Introduction 1 1 Unmanned Marine Vehicles 0 4 1 1 1 Underwater Vehicles 04 1 1 2 Surface vehicles gt sin a eee Be RY 1 2 Outline of This Thesis 04 The vessel 2 1 How Does A Sailboat Work 0 004 DD MOOS ctxt AE AR tus Sep aw dees Dees a 2 2 1 Sail Construction 0 2 00000 085 2 2 2 Sail Attachment 2008 23 A PIRMA Sailboats Bones a foo ae ae A e State of Art 3 1 Self Steering Gear a 3 1 1 Mechanical Self Steering 3 1 2 Electronic Self Steering 3 2 Automatic Sail Control 0 3 3 Recent Technical Innovations 3 4 Autonomous Sailboat Competitions 3 5 Field Applications 2 2 2004 Navigation And Control 4 1 Low Level Control 0 0 000000 eee 4 2 Autonomous Sailboat Navigation Onboard Electronics 5 1 System Hardware 000000 0b eee 5L Waspmote selg romana RA eR eke ee ds 5 1 2 Electrical Characteristics Bld Input Output sedal 6 2 8 Sous Hees ee Ree eat 5 1 4 Working Environment 10 11 12 13 14 17 17 17 18 19 20 23 30 31 32 35 ii CONTENTS 5 1 5 Over The Air Programming OTAP 45 5 2 RF Radio Module 0000000008 46 5 3 Electr
7. designs developed at Aberystwyth They are robust yet light sail designs whose wing like profile is promoted as aerodynamically more efficient that the classical soft sails On the downside rigid sails demand a careful and expensive design based on high tech materials they can not be reefed and are considered fragile 3 3 RECENT TECHNICAL INNOVATIONS 21 under bad weather conditions On the other hand the balanced rig is self tacking reduces mechanical loads on the rig and overall can be more en ergy efficient than an equivalent sloop rig 11 Figure 3 2 Balanced rig concept 10 Flexible hull Protei 12 is an international volunteer based organi zation that is designing a new concept of sailboat that can be applied on autonomous cleaning of oil spills In this kind of application the boat must tow a large floating structure designed to absorb the oil at surface This structure renders useless the classical stern rudder to head the boat To deal with this basic problem a flexible shape shifting hull has been pro posed by the Protei team According to the authors its benefits are 22 CHAPTER 3 STATE OF ART e Better trajectory control greater pulling capacity e Better maneuvering smaller radius of turning dynamic stability e Lateral lift due to curved hull profile at high speed e Less resistance and turbulences no centerboard no rudder compen sating environmental noise with flexible body 8
8. heading information and instantaneous pitch and roll angles over a RS232 interface The board temperature and raw readings from three magnetometers can be also obtained The compass readings are tilt and roll compensated till 50 degrees The TCM2 50 can not operate for heeling or pitching angles over that limit The TCM2 50 has a maximum update rate of 20Hz It is powered at 5 V and consumes 20 mA This board is connected to one of the microcontroller TTL serial ports using a simple level converter circuit based on a MAX3232 integrated circuit IC The MAX3232 36 is an IC first created by Maxim Integrated Products that converts signals from a RS 232 serial port to signals suitable for use in TTL compatible digital logic circuits and vice versa The MAX3232 is a dual driver receiver and typically converts the RX TX CTS and RTS 48 CHAPTER 5 ONBOARD ELECTRONICS signals The receivers reduce RS 232 inputs which may be as high as 25 V to standard 5 V to 3 V TTL levels These receivers have a typical threshold of 1 3 V and a typical hysteresis of 0 5 V In Figure 5 8 we can see the connection diagram to connect the TCM2 50 to the Waspmote board 5w TEM 2 50 3 oy 1345 C1 ce Aus Serial 1 Ax Waspmote Aun Serial 1 Tx Cd Figure 5 8 TCM adapter 5 4 GPS Receiver The Waspmote board is prepared to accept an A1084 Figure 5 9 20 chan nel GPS receiver with an external antenna This receiver is based on the SiRF
9. is displayed on the serial monitor through the USB connection DEBUG TELE This directive permits debugging the telemetry packets Output is displayed on the serial monitor through the USB con nection DEBUG TIME This directive activates the profiling sen tences embedded in the code to know the time consumed in the main blocks of the control code Measurements are displayed on the serial monitor DEBUG TIME SD This directive is complementary of the DEBUG_TIME directive and when de fined it makes the profiling data to be saved on the SD card 71 72 CHAPTER 6 CONTROL SYSTEM DEBUG MEM This directive activates memory checking functions that are invoked at selected loca tions in the code to assure memory consis tency Information gathered is displayed on the serial monitor The control application is structured around the following modules e Initialization e Fuzzy Logic System Rudder Control Sails Control e Control loop Remote control mode Autonomous mode e Robust radio connectivity Send Receive e Sensor sampling e SD Card Full sensors message Timing measurements message The flowchart of the software on board is detailed on Figure 6 4 6 2 SAILBOAT CONTROL SYSTEM 73 Initializatiom Yes Any new package Yes onnect to base station Waypoint Home Yes Remote control
10. option The next autonomous message will send the calibration command When the sailboat receives the command it will calibrate the sensors The calibration will be stored on the EEPROM Table 6 11 Calibrate sensors 67 68 CHAPTER 6 CONTROL SYSTEM Name Set sailboat Id UCI emission rate Actor User Description To set emission rate of the sensors message Trigger Set the emission rate option Result The emission rate will be set on the sailboat Flow To set the emission rate The next autonomous message will send the emission rate command When the sailboat receive the command it will set the emission rate Table 6 12 Set sailboat emission rate Name Set navigation Id UC10 execution rate Actor User Description To set navigation execution rate Trigger Set the navigation execution rate option Result The navigation execution rate will be set on the sailboat Flow Set the emission rate The next autonomous message will send the navigation execution rate command When the sailboat receives the command it sets the navigation execution rate Table 6 13 Set navigation execution rate 6 2 SAILBOAT CONTROL SYSTEM 69 Name Set sailboat Id UC11 mode Actor User Description Sets the sailboat mode Trigger Push C on the wirel
11. radio modules have been pro grammed to resend automatically dropped or incorrect radio packets for a number of times The loss of telemetry packets is not critical because they are still logged on the micro SD card available on board More critical is the loss of command packets sent from the base station and this is why these packets must be acknowledged explicitly from the sailboat Send The sailboat sends to types of packages telemetry packages and waypoint reception acknowledgment packages Telemetry messages have the format that described previously in Table 6 1 Waypoint acknowledgment packages are just copies of the message received from the base station and described in Table 6 2 Receive When the sailboat receives a package it is parsed and the message type detected Unrecognized messages are silently ignored The vessel receives two different messages e Autonomous mode packages Described on Table 6 2 e Remote control mode packages Described on Table 6 3 86 CHAPTER 6 CONTROL SYSTEM 6 2 6 Sensor Sampling Sensors are sampled at different rates depending on its potential rate of change and the temporal cost of a new reading in order to reduce the mean sampling time and hence the duration of a control cycle The GPS sensor available on board has a maximum update rate of 1 Hz and it does not make sense to read it faster because it will deliver old estimates Even more while reading at the nominal rate of 1 Hz ti
12. several other sensors for measuring for example the board temperature or the battery level 39 40 CHAPTER 5 ONBOARD ELECTRONICS This board is prepared to accept external hardware modules like a GPS receiver a GSM GPRS modem or different XBee RF communication mod ules It is powered from a 3 7V 6000 mAh Li Ion battery and consumes 9 mA under normal operating conditions Suitable photovoltaic panels can be connected directly to the board to recharge the main battery 5 1 1 Waspmote The Waspmote board is an embedded system inspired in the microcontroller based electronic prototyping platform Arduino 32 but Waspmote gives us more sensors and sockets in only one board Its hardware architecture is modular in a way that we can add or remove off the shelf hardware modules easily to the device The modules available for integration in Waspmote can be categorized e ZigBee 802 15 4 modules 2 4GHz 868MHz 900MHz e GSM 3G GPRS module Quadband 850MHz 900MHz 1800MHz 1900MHz e GPS module e Several sensor modules sensor boards e Storage Module micro SD Memory Card e Photovoltaic solar panels Figures 5 1 and 5 2 show the main components and sockets available for connecting external devices and power sources 5 1 SYSTEM HARDWARE Al Microprocessor GPRS Accelerometer 12C UART Socket Expansion Radio Board Socket XBee Sockets Sensor I O Battery Socket Crystal Oscillator RSSI Leds Reset But
13. the station awaiting the restitution of radio communications Retum To Home Sailboat Position Retum To Home i 15 4283 15 4282 15 4281 15 428 15 4279 15 4278 Longitude Figure 7 8 Return To Home 15 4277 104 CHAPTER 7 RESULTS 7 3 3 Excessive Heeling Effects The A TIRMA exhibits a behavior that is common to many sailboats when they are sailing close hauled and they heel in excess Under those conditions the observable behavior is a clear tendency to luff up or haul upon wind loosing the bearing completely To avoid this type of instability it is not only necessary to avoid extreme heeling angles but also compensate the tendency to luff up with the rud der To neutralize this effect the rudder position computed by the fuzzy controller is modified when a heeling angle threshold is reached The rudder correction is directly proportional to the heeling of the sailboat It is worth noting that this kind of maneuver is routinely done by helmsmen on board of full scale sailboats Figures 7 9 and 7 10 depict the consequences of not compensating for this effect In Figure 7 10 it is evident how abrupt bearing changes follow once extreme roll i e heeling angles are reached All this end up in a unstable course of polygonal appearance see Figure 7 9 characterized by close hauled course segments followed by hauls upon the wind lose of speed tack for bearing correction and so on continuously
14. 5 1 1 5 2 2 5 3 3 5 4 Cycles x10 Figure 7 13 Servo Battery 7 3 NAVIGATION TASKS 109 Waspmote Battery 60 55 50 45 40 35 Batery 30 25 20 0 0 5 1 1 5 2 2 5 3 3 5 4 Cycles x10 Figure 7 14 Waspmote Battery We can see in both graphics that in three and a half hours the wasp mote consumed only 10 of battery s capacity and the voltage of the servo battery decreased in 0 5 volts Note that the minimum operating voltage of servos is 4 volts 110 CHAPTER 7 RESULTS Chapter 8 Conclusions This work has been motivated by the necessity of developing an small and affordable autonomous sailboat platform that could be transported and op erated by one or two people without any special means In agreement with those objectives this master thesis has described the design of a low cost autonomous sailboat whose development has been based on a standard RC One Meter class vessel and off the shelf low power hardware components The main features of the described system are its flexibility as exper imental platform its large power autonomy and its robustness in case of communication failures Extensive sea trials in autonomous mode have been performed to confirm the stability and robustness of the A TIRMA control system The sailboat in spite of her dimensions has exhibited seaworthiness and demonstrated to be an interesting experimental platform The amount of space and displacement available in a One Meter
15. 6 9 6 10 6 11 6 12 6 13 6 14 7 1 7 2 Oceanographic platform relative capabilities in normal operating condi tions taken from 16 lt o e yd ee a o es FAST University of Porto FEUP aoaaa BeagleB Aberystwyth University o so oo a e e Gill the Boat US Naval Academy osoo a Avalon Swiss Federal Institute of Technology Zurich ETH Roboat I Austrian Society for Innovative Computer Sciences INNOC VAIMOS cole Nationale Sup rieure de Techniques Avanc es Bretagne ENSA ao ty it ation Seeds Best E ee SER whe as iia ee EN Variable meaning 1 e Flowchart symbols eine ke fsa dk eee Se ko a en ee a Aa v GPS receiver features ee a a a a e Power demands of system components Format of telemetry messages 1 eee ee ee Format of an autonomous mode message Format of a control message 1 ee Greate waypoint verd hoe ee al a a a aa Up way pom saree Bo Ste is om At a eae A A aw ee Down waypoint lt lt a fie She ies PG a ee ee ee a Adee Delete waypoint s 6 32a sos ie ee ee a ee Bhs Bs Send autonomous message soso soos e Send control message aoyo asara SS eee A Sees eg he ee Receive sensors message 1 6 ee e Callbrate sensOrS ce da A Soe ee Re ae Set sailboat emission rate Set navigation execution rate S
16. 8 113240 28 113241 0 1112m A 4 AVR Libc Error We saw that the avr libc 52 version used in Waspmote API had a bug on the management of the dynamic memory This error appeared when we get and free memory during a lot of time It produced fragmentation of the memory when trying to free acquired memory the problem was that really the mem ory was not being freed So we lost memory because avr libc s functions malloc and free did not work correctly To solve the problem we adapted the solution for Arduino to our system We used a new implementation for memory management developed by Joerg Wunsch 53 and we placed it in our API directory 118 APPENDIX A IMPROVEMENTS ON WASPMOTE API Bibliography 6 7 8 9 ARGOS Program Information Center http wo jcommops org cgi bin WebObjects Argo Autonomous Underwater Vehicle http en wikipedia org wiki Autonomous _underwater _vehicle Autonomous Undersea Vehicle Apllication Center Portal on AUV technology http auvac org Underwater gliders http en wikipedia org wiki Underwater_ glider Autonomous Surface Vehicle http en wikipedia org wiki Unmanned _surface _vehicle Liquid Robotics http liquidr com technology wave glider html Global Ocean Drifter Program http www aoml noaa gov phod dac index php SkySails GmbH Web site http www skysails info english Roberts G Trends in marine control systems Annual reviews in control 32 2 263 26
17. 8 1323 28 1322 28 1322 28 1321 loa 15 4283 15 4283 15 4282 15 4282 15 4281 15 4281 Longitude Figure 7 7 An accuracy test The red line it is completely a sidewind course and we can see that the sailboat kept the course till it arrived to the waypoint Then in green the sailboat started a downwind course that turned into a sidewind course And finally the blue line is a upwind course and we can see how the sailboat do a series of tacks to reach the initial waypoint Table 7 4 Accuracy task JK length 31 48 m KL length 21 43 m LJ length 37 74m Latitude 28 1326 28 1325 28 1324 28 1323 28 1322 28 1321 28 132 28 1319 15 4284 7 3 NAVIGATION TASKS 103 7 3 2 Return To Home RTH The RTH it is an important behavior of the sailboat because it guarantees that if the sailboat loses the connection with the base station it will initiate automatically the return to the Home waypoint On Figure 7 8 we can see how the sailboat automatically returns to home after losing the radio link The sailboat was performing in a triangular route drawn in blue After one round we disconnect the radio antenna at the base station when the sailboat was still close to the upper vertex After 20 seconds without any reception from the base station the sailboat changed the active waypoint to the Home waypoint initiating the trajectory depicted in red After reaching the Home waypoint it tries to keep
18. 9 2008 http 202 114 89 60 resource pdf 2236 pdf 119 120 BIBLIOGRAPHY 10 BalancedRig BalancedRig website Available online http balancedrig com description html last accessed Julyl 2013 2013 11 Pigi Stelzer R Estarriola Dalmau D A Study on Potential Energy Sav ings by the User of a Balanced Rig on a Robotic Sailing Boat in Pro ceedings of International Robotic Sailing Conference pp 87 94 Cardiff Wales UK 2012 12 os Protei Open Source Sailing Drone Web site http www protei org 13 Sailbot competition website http sailbot org 14 World Robotic Sailing Championship web site http www roboticsailing org 15 Sailbuoy web site http sailbuoy no 16 Rynne P F von Ellenrieder K D Unmanned autonomous sailing Current status and future role in sustained ocean observations Mar Technol Soc J vol 43 no 1 pp 21 30 2009 17 Rynne P F von Ellenrieder K D Development and Preliminary Ex perimental Validation of a Wind and Solar Powered Autonomous Sur face Vehicle IEEE JOURNAL OF OCEANIC ENGINEERING VOL 35 4 pp 971 983 October 2010 18 Pith Stelzer R Jafarmadar K History and Recent Developments in Robotic Sailing in in Proceedings of International Robotic Sailing Con ference pp 3 23 L beck Germany 2011 19 Stelzer R Jafarmadar K The Robotic Sailing Boat ASV Roboat as a Maritime Research Platform in Proceedi
19. Collision safe 5 Remotely controlled Highly visible Pa gt to become autonomous 10 Green Affordab hae 6 Upwind sailing ue i 9 Sensors controlled Figure 3 3 Protei a shape shifting hull sailboat Energy saving Most of the energy consumed at a sailboat is used in sheeting sails and in general energy aware control systems do not adjust the sails except when necessary A good example of this approach taken to the extreme that is no sail actuator is used is Sailbuoy 15 In this 2m boat the wing sail is free to swing from side to side following the wind and its angle is not regulated Accordingly no wind sensor exists on board Sailbuoy is also remarkable because it is to our knowledge the only com mercial product in this ambit designed to act as a surface sampling vehicle 3 4 AUTONOMOUS SAILBOAT COMPETITIONS 23 Figure 3 4 Sailbuoy 15 3 4 Autonomous Sailboat Competitions Two international competitions Sailbot 13 in North America and the World Robotic Sailing Championship 14 in Europe are common meeting forums for researchers and students working this field It is beyond the scope of this document to extensively review all sail boats that have participated on this competitions and that information can be obtained from the competitions web sites The following tables just pro vide some examples of autonomous sailboats that have taken part in the last editions of the Worl
20. III chipset and supports the NMEA0183 37 and SiRF binary serial protocols We use the binary protocol to configure the receiver elevation mask signal strength mask messages rates and rely on NMEA RMC and GGA messages for obtaining information about position altitude hor izontal dilution of precision hdop ground speed course and time It has 5 4 GPS RECEIVER 49 a nominal accuracy of less than 10 meters However in our tests at sea imposing elevation and signal strength masks the accuracy has been quite stable and typically better that 3 meters It is powered by the on board 3 3 V regulators and consumes 26 mA 38 In the Table 5 1 we can see the main characteristics of the GPS system Table 5 1 GPS receiver features Model A1084 Vincotech Movement sensitivity 159dBm Acquisition sensitivity 142dBm Hot Start time lt ls Warm Start Time lt 32s Cold Start Time lt 35s Antenna connector UFL External Antenna 26dBi uel ESA WASPMOTE GPS ofjgmmo Figure 5 9 GPS receiver and external antenna 50 CHAPTER 5 ONBOARD ELECTRONICS 5 5 Wind Vane The wind vane has been custom built from an Optimist wind vane attached to a US Digital s MA3 miniature absolute magnetic encoder 39 which is shown in Figure 5 10 The encoder is installed in an aluminum enclosure with a floating cap on top of the mast and connected to one of the analog inputs of the microcontroller Figure 5 11 It allows to detec
21. R 1 INTRODUCTION the acquisition of data series with the desirable temporal and spatial density 1 1 Unmanned Marine Vehicles The necessity of cutting costs of data gathering and sampling for longer periods is pushing forward the development and utilization of unmanned vehicles in oceanographic campaigns Unmanned marine vehicles can be broadly classified into two different categories underwater vehicles and sur face vehicles 1 1 1 Underwater Vehicles This category includes profilers propelled underwater vehicles also known as AUVs Autonomous Underwater Vehicles underwater gliders Profilers and floats are devices capable of traversing the column of water regulating their buoyancy They can only control its motion in depth and in the long term they moved in the ocean carried away by marine currents Their power economy makes them very interesting instruments for long de ployments The ARGOS program 1 is perhaps the best known example of a large scale program based on this kind of devices An AUV 2 is a propelled torpedo shaped robot which travels underwa ter autonomously Their range and payload capacity is basically limited by its dimensions Numerous examples of this kind of vehicles can be found on 3 AUVs are now being used for more and more tasks with roles and mis sions constantly evolving Meaningful cases of use can be found in science missions and commercial tasks in the oil and gas industries which use AUVs t
22. UNIVERSIDAD DE LAS PALMAS DE GRAN CANARIA Master Oficial en Sistemas Inteligentes y Aplicaciones Num ricas en Ingenieria DE LAS PALMAS I NI ARIA SINANI Master Thesis A TIRMA A small autonomous sailboat ngel Ramos de Miguel Tutores Jorge Cabrera G mez Antonio Carlos Dom nguez Brito 29 July 2013 Agradecimientos A toda mi familia por todo el nimo que me han dado en este trabajo asi como en todos los aspectos de la vida Por prestarme atenci n y entenderme cada vez que les explicaba cualquier aspecto del proyecto Al todo los que trabajan en el instituto universitario de sistemas in teligentes y aplicaciones num ricas que me han hecho disfrutar de un a o adquiriendo conocimientos que seguramente utilice el resto de mi vida pro fesional Por supuesto a mis tutores D Antonio Carlos Dom nguez Brito por pro ponerme el proyecto y toda la ayuda prestada en todas las fases y a D Jorge Cabrera G mez por todas las horas d as y meses que ha dedicado en explicarme y ayudarme durante todas las fases del proyecto No puedo olvidar a los compa eros del equipo que han participado y espero que participen durante mucho tiempo a Jos Daniel Rodr guez Sosa y a Bernardino Valle Fern ndez por las horas de dise o debates de como deber a comportarse el barco y las horas en la playa haciendo pruebas Por ltimo a mi novia por todas las horas que me ha esperado estando yo en el laboratorio Abstra
23. a deg Figure 4 2 Desired heeling function 23 34 CHAPTER 4 NAVIGATION AND CONTROL The optimal heeling function depends on the variables detailed on Table 4 1 Table 4 1 Variable meaning Rimas Maximum heeling of the sailboat v Actual velocity of the sailboat Umor Maximum velocity of the sailboat Q Wind direction k boat specific constant determined by experiments a Too low Optimal Too high Pp v LD vo 205 S o 2 2 a 0 50 50 heeling deg a Ten Keep Ease off 2 D Q E o 205 Q 2 O o o0 j 10 5 5 10 Change pct of max Figure 4 3 Fuzzy sails 23 The fuzzy sets included in Figure 4 3 define the vocabulary for the desired heeling Using the values given by the function shown in Figure 4 2 the actual heeling of the sailboat is subtracted to the desired heeling and the difference is used to calculate the new sails position with the fuzzy sails 4 2 AUTONOMOUS SAILBOAT NAVIGATION 35 control rules 4 2 Autonomous Sailboat Navigation In 30 it is proposed an algorithm to calculate the next best bearing for a sailboat The basic idea behind the algorithm is to select the bearing that maximizes the velocity made good or velocity projected in the direction to destination This basic idea is modulated by a hysteresis parameter to avoid tacking constantly The algorithm uses as basic parameters the real wind direction and speed the destination coordinates
24. a shape shifting hull sailboat 7 22 Sailbtloy 15 i hos he a ve he let Mead Aig Bed seat Shh ae eee io ed 23 FAST HHUP por opia Pa as el ee ed ee 24 Beagle Aberystwyth sooo ee 25 Gill The Boat USNA ue 2 God Se amp ah koe Md ee he as 26 Avalon Swiss Federal Institute of Technology Zurich ETH 27 Roboat I Austrian Society for Innovative Computer Sciences INNOC 28 VAIMOS Ecole Nationale Sup rieure de Techniques Avanc es Bretagne ENS E tt ae A She Bae ae ep BO 29 Fuzzy rudder position 23 2 ee 32 Desired heeling function 23 2 2 ee 33 Fuzzy sails 23 aia a bot ae eee ae ee a Pe a th ee A 34 Bearing selection algorithm 30 36 Waspmote board V1 1 Topside 0 048 41 Waspmote board V1 1 Bottom side 41 Waspmote pinout vai a a a 43 Waspmote IDE ota ai a as e RS Sek gk Rove oP 44 Waspmote IDE Buttons 00084 44 XBee Module s e aaa Ae AA ee eee eh as 46 The 868 MHz OdBi antenna in its final placement 47 TOM adapter so Soit a Ba eA be a i a eaa 48 GPS receiver and external antenna e 49 MA3 absolute angular encoder o oo ooa 0000 50 Home made wind Vane 50 vi 5 12 6 1 6 2 6 3 6 4 6 5 6 6 6 7 6 8 6 9 6 10 6 11 6 12 7 1 7 2 7 3 7 4 7 5 7 6 7 7 7 8 7 9 7 10 7 11 7 12 7 13 7 14 LIST OF FIGURES AGS 12 o hates 3 Ge hoes otter rr of deta
25. acement 25 kg Draft 1 5 m Beam 0 6 m Hull Type Custom designed hull built for the SailBot class Sail Type Bermudan style fabric mainsail and jib Sensors and Computers Rabbit controller mounted on a in house designed and built board that includes a GPS receiver Relative wind is measured by an optical encoder and she has a working ultrasonic collision avoidance system Other notes She was wholly designed built and is operated by the midshipmen of the United States Naval Academy She has reached speeds of over 6 knots and has taken voyages up to 21 nautical miles Figure 3 7 Gill The Boat USNA 3 4 AUTONOMOUS SAILBOAT COMPETITIONS 27 Table 3 4 Avalon Swiss Federal Institute of Technology Zurich ETH Team Swiss Federal Institute of Technology Zurich Location Ramistrasse Ziirich Swiss Class Microtransat Length 3 95 m Hull Type A monohull design Sensors and Computers The control system is implemented on a MPC2120 industrial computer running Linux Power Systems The power supply is realized with four solar panels of 90 Wp four lithium manganese batteries of 600 Wh each and a direct methanol fuel cell for back up power Figure 3 8 Avalon Swiss Federal Institute of Technology Zurich ETH 28 CHAPTER 3 STATE OF ART Table 3 5 Roboat I Austrian Society for Innovative Computer Sciences INNOC Team Austrian Society for Innovative Computer Sciences Location Viena Austria Euro
26. and makes for a smooth and clean entry where the wind meets the sail The luff hem method eliminates scalloping which can happen with the tubes and it distributes the wind load evenly over the entire luff The sail tubes provide for excellent swiveling as the sail tacks but produce point loads where they attach to the sail and this can cause the sail to wear out prematurely The head of the jib usually contains a grommet to which is attached a halyard For ease of adjustment the halyard is fixed usually up at the jibstay attachment point Tension is adjusted on the sail luff by use of a 14 CHAPTER 2 THE VESSEL downhaul which is tied to the tack grommet led to the jib club and back about half way down the club to some sort of a tension adjustment like a cleat 3 hole bowsie or other device The clew is constrained in the vertical direction by something as simple as a loop of sheet line tied through the clew grommet and led under and around the club A second line called the clew outhaul leads aft to the end of the jib club and provides for adjusting the position of the clew along the club controlling the fullness of the foot and the bottom third of the jib The corners of the mainsail are rigged in the same way More details on model sailboat rigging techniques can be found on 28 2 3 A TIRMA Sailboat The sailboat described in this master thesis has been named A TIRMA or Autonomous TIRMA after the famous canary TIRMA sa
27. and the sailboat position and polar Figure 4 4 shows a flowchart of the bearing selection algorithm This algorithm is considered by the authors as a short term routing algorithm In fact that is possible as far as the following two conditions hold e The true wind is the same all over the area between the current boat position and destination e The true wind will remain for the whole leg as it is in the moment The flowchart starts with the initialization of the variables that the func tion will use The data needed to select the next bearing are Table 4 2 Flowchart symbols B Current boat position T Target position pvp Current boat heading y W True wind direction and speed n Hysteresis factor The current boat position and orientation are needed to decide if it is necessary to do a tack maneuver then the wind speed it necessary if the 36 CHAPTER 4 NAVIGATION AND CONTROL initialization B T 9 0p v t T B Pu abs inv P W ars right hand side optimum Ve R max Vir Pb max R Pw abs inv Be i ad ee ee je left hand side in analogy to right hand side i optimum i Von for W abs Pu absin amp i 1 1 1 determining V max aNd Py max R or L closer to current direction hysteresis V yes new direction Pb new i Ph max L 1 right hand side optimum closer to current boat direction 2 left hand side optimum closer to current boat direct
28. any kind of profiling support Hence timing 91 millizecodz 92 CHAPTER 7 RESULTS data have been obtained inserting timing functions in the code as explained in detail in the previous chapter This approach certainly degrades full cycle measurements like the cycle time because these measurements will be distorted by the cost of invoking the timing functions themselves but it is to our knowledge the only viable approach Times I T T 14000 T Non Optimized Optimized 12000 10000 8000 5000 4000 2000 ha fi N 7 a Le K PAS af trol y NA i U 9 wy n Joe y H 20 Y A V y A a 0 200 400 600 800 1000 1200 1400 1600 Figure 7 1 Timing with and without code optimizations We can see in Figure 7 1 that the non optimized sometimes had huge time peaks the biggest of 12409 milliseconds We analyzed the data looking for the culprit functions that incurred in these delays It was discovered that the functions used to send data through the XBee radio module could get locked under certain circumstances The data analysis also showed that the function used for getting the GPS position could incur in occasional delays of 1 to 3 seconds We rewrite some parts of these Waspmote s API functions to reduce the time of each delay to a maximum of 100 milliseconds 7 1 CODE IMPROVEMENTS 93 In the first versions of the on board control system the cost of logging on the SD card was one
29. ard inclinometers and wind vane zero must be recalibrated In this mode the base station sends every 5 seconds short messages to check the state of the radio link If these packets are not received at the vessel for 20 seconds the active waypoint is cleared and substituted by the coordinates that identify the Home Point and the sailboat tries to arrive to that point autonomously This constitutes the Return To Home or RTH behavior that has proved a valuable capability during sea tests This situation can be reverted from the base station as soon as the radio link is reestablished In that moment new waypoints and navigation parameters can be transmitted to the boat RC Control command The format of RC control commands is CTrxrstzx An RC control command command is sent every 500 milliseconds from the base station to the vessel if the system is in radio control mode This 6 1 BASE STATION 63 message specifies the position of the rudder in degrees and the percentage in which the sails must be tightened as detailed in Table 6 3 Table 6 3 Format of a control message C Indicates that it is a control command TLL Indicates the desired rudder position SLL Indicates the desired sails position 6 1 8 Use Cases This section defines the different cases of usage of the base station 64 CHAPTER 6 CONTROL SYSTEM Name Create way Id UCI point Actor User Descri
30. are silent vessel with a virtually unlimited operational range restricted only by wind and sea conditions that may host large pay loads and measure atmospheric and ocean parameters simultaneously All these features confer them an enormous potential to serve as sampling and monitoring marine vehicles All in all the field has still to mature and produce reliable and ro bust designs capable of dealing with hard conditions at sea on their own means Regardless of this state of affairs field applications developed with autonomous sailboats have been published Here we summarize some of them e Acoustic mapping and cetacean tracking Recently the IN NOC s ROBOAT sailboat has been used to acoustically track cetaceans in the Baltic Sea 19 under strong sea conditions e Water quality sampling measurements Rynne and von Ellen rieder 17 have described the development of the Wasp a 4 2m mono hull keelboat with wing sail and its application in seawater quality control Chapter 4 Navigation And Control This master thesis has focused on the development of the low level control of the rudder and sails and a bearing selection algorithm i e the determination of the optimal bearing to navigate the sailboat to a destination taking into account the wind direction This work has been based on two influential and recognized articles of the field Fuzzy logic control system for autonomous sailboats 23 and Autonomous sailboa
31. as six positions The first one the zero posi tion gives the location of the Home waypoint The other five points are the waypoints that define the route and are used by the bearing selection algorithm The bearing selection algorithm 30 is used at the next control level to obtain the best i e fastest bearing to reach the active waypoint given the current wind direction boat position and heading This control level runs at an adjustable frequency but can be triggered also by a sudden wind roll Note that as far we lack an estimation of wind speed on board we run this algorithm using only the apparent wind The algorithm reacts in real time to wind changes like a sailor does The parameters needed for the calculation of the desired boat heading are e Target position e Current boat position 6 2 SAILBOAT CONTROL SYSTEM 83 e Current boat heading e Apparent wind direction The bearing selection algorithm is the slowest function on the on board system roughly 150 milliseconds and it does not make sense to calculate the goal bearing on every iteration that is each 250 milliseconds Thus the bearing selection function is run by default every five seconds although it can be modified using the base station application or if the wind direction has changed Note also that the execution of the bearing selection function will be delayed if the sailboat is turning in that moment because wind vane readings would be unreliable The
32. ast control mode a wireless game pad is used to control the rudder and the sails 2500 MapTracker Pitch Koll xBee xBee TCM Boat Station Acc X 602 i 502 E 400 e Acc Y 302 z 202 E AGG Z WULZS 13Z11M 15 4Z834AUEJUUUNIUUUCU A 10 ul 256 3M 15 452498C74 5P 1 1R 4 5T24 50W147X 29Y 212Z Local xBee Power 4 Received WOL28 13211M 15 42854A0E5000N5000C0 Battery Send i WOL28 13211M 15 42854A0E5000N5000C0 Wasp u Rece Servos 1l WOL28 13211M 15 42854A0E5000N5000C0 WayPoints Distance Boat Emition Time PC ees J wo lt gt W1 94 80745 5000 6 W1 lt gt W2 4022 E 2 W1 lt gt 80 8 E E7 W2 lt gt W3 82 44316 Boat Nav nme Last Cycle Time 5000 O ms Max Rudder Roll Rudder Factor ran PP IDrate 20 fos Pitch PE ERE Polar Limit Up Polar Limit Down 50 160 Roll ES OR C Mode Connect Figure 6 1 Base Station GUI 6 1 BASE STATION 55 6 1 1 Base Station Hardware To use the base station application we need a XBee gateway and a wireless game pad to communicate with the sailboat and to command the boat in RC control mode As XBee gateway we have selected the one provided with the Waspmote Figure 6 2 XBee gateway The selected wireless game pad was a Logitech WingMan Cordless 56 CHAPTER 6 CONTROL SYSTEM Figure 6 3 Logitech WingMan Cordless 6 1 2 Software Architecture The application receives and trans
33. bearing selection algorithm described in 30 is based on absolute wind direction In our sailboat we only have a wind vane so we only have the apparent wind Figure 6 12 illustrates the difference between the true wind and the apparent wind in the case where the sailboat speed is 1m s and the wind speed is 3m s Considering the apparent wind direction as the absolute wind direction provokes that the boat will never sail as close hauled as it could as deter mined by its polar In downwind courses the situation is just the opposite the vessel can go into a too closed downwind course inadvertently At the lowest level the fuzzy controller described previously see 6 2 3 runs at the highest frequency of the control system and controls the sail and rudder positions to keep the sailboat under control on the desired bearing This controller has been implemented using the EFLL library 45 The sailboat can transit into autonomous mode if a prolonged failure of radio communications is detected or because this control mode is ex plicitly commanded from the base station through a radio packet with the format WaLlaxMaxArxE xx The preamble W identifies this packet as an autonomous mode command packet the L and the M fields indicate the latitude and the longitude of the active waypoint A indicates how far in 84 CHAPTER 6 CONTROL SYSTEM true wind app wind error Figure 6 12 Apparent wind error meters can be the
34. by Sperry Gyroscope Company in 1911 21 This system compensated for varying sea states using a feedback control and automatic gain adjustments 3 2 AUTOMATIC SAIL CONTROL 19 Nicholas Minorsky did a contributions to autonomous ship steering he presented a detailed analysis of a position feedback control He formu lated the specification of a proportional integral derivative PID controller n 22 But this type of controller has two disadvantages 1 It is difficult to adjust it manually because the operator usually does not have enough insight about control theory 2 The optimal adjustment varies depending on the situation Changing circumstances require readjustment of settings Due to the dynamic and ever changing environment artificial intelligence and fuzzy logic has been studied for rudder control Various publications have shown the suitability of fuzzy logic for rudder control and also for sail control 23 Polkinghorne et al 24 did a comparison between a conven tional PID and a fuzzy logic controller The experiments shown a much smoother rudder action for the fuzzy logic controlled rudder 3 2 Automatic Sail Control Most sail control strategies published for autonomous sailboats are based on measuring the apparent wind Many of them have virtually an infinite number of sail positions when the system sets a sail position the real po sition depends of the resolution of the actuator Some researchers use onl
35. calibrated offsets are valid These offsets are stored at fixed addresses in EEPROM At the conclusion of this stage the sailboat will be in remote control mode and it will start sending telemetry data through the radio every 5 seconds by default 6 2 3 Fuzzy Logic System The sailboat s actuators rudder servo and sails winch are controlled using a fuzzy control system 23 Setting this kind of controller involves defining the control vocabulary and associated membership functions for inputs and outputs and the rules to control the rudder and the sails The fuzzy 6 2 SAILBOAT CONTROL SYSTEM 75 controllers used in the A TIRMA has been inspired in those described in 23 even though they differ in important aspects Rudder Control The input data for the rudder control circuit are the current boat direction and the desired direction from the routing system The difference between these two gives the necessary course correction which enters directly into the fuzzy system as input variable Figure 6 5 Also in order to avoid over steering we use the position of the rudder Figure 6 6 as an additional input variable Finally we have a output fuzzy set with the differents posible positions of the rudder Figure 6 7 The definition of membership functions have been adapted from 23 to accommodate the characteristics of our sailboat Our fuzzy sets for the rudder control are Input desired direction Strong Left Left Neutra
36. ct This master thesis describes the development of the A TIRMA an au tonomous sailboat based on an One Meter Class RC vessel The A Tirma navigates autonomously with no human intervention except for establishing the route it will follow One of the distinctive features of sailing robots is that they constitute really interesting operative platforms for monitoring and data sampling in aquatic environments mainly due to its low power consumption requirements which allows them great operative autonomy In addition in the case of A TIRMA it is also easily deployable due to its small size and weight The control system of A TIRMA consists of two main parts a base sta tion and the sailboat From the base station a notebook we can command straightly the sailboat s sails and rudder remotely in RC Remote Control mode using a wireless link In autonomous mode we can make the boat to follow autonomously a route which is a list of waypoints specified on a Google Maps interface The boat is endowed with an embedded on board system which in autonomous mode bases its operation in an autonomous fuzzy logic based navigation algorithm for controlling the boat s sails and rudder in order to navigate autonomously following the route of waypoints specified from the base station During its development we have made in tensive sea trials to check the operational capabilites of A TIRMA and its control system at sea and its autonomy in terms of power con
37. d Robotics Sailing Championship 46 to provide the reader a summary on the main characteristics of the boats CHAPTER 3 STATE OF ART Table 3 1 FAST University of Porto FEUP Team FAST University of Porto FEUP and INESC TEC Location Porto Portugal Class Microtransat Length 2 5 m Displacement 60 kg Draft 1 25 m Beam 0 67 m Hull Type Based on a scaled down version of a Class 40 hull Sensors and Computers FPGA based computer running Linux Power Systems 190 W h of Lithium Ion batteries and 45 W peak of photovoltaic solar panels Figure 3 5 FASt FEUP 3 4 AUTONOMOUS SAILBOAT COMPETITIONS 25 Table 3 2 BeagleB Aberystwyth University Team Department of Computer Science Aberystwyth University Location Aberystwyth Wales Uk Class Microtransat Length 3 65 m Displacement 280 kg Draft 0 65 m Beam 0 86 m Hull Type Based on a Mini J disabled sailors dinghy hull Sail Type 2 square metre carbon fibre wing sail Sensors and Computers GPS Fluxgate Compass Ultrasonic windsensor Gumstix Single Board Computer Power Systems 2 8 KW h of lead acid batteries and 90 W peak of photovoltaic solar panels Communications Wifi Iridium Satellite Modem optional GSM i Figure 3 6 Beagle Aberystwyth 26 CHAPTER 3 STATE OF ART Table 3 3 Gill the Boat US Naval Academy Team US Naval Academy Location Annapolis Maryland USA Class Sailbot Length 2 m Displ
38. d control system due to RAM memory restrictions GPS Accuracy Longitude Figure 7 5 GPS Accuracy 7 3 Navigation Tasks In order to test some of the navigation capabilities of the sailboat we have taken inspiration from tasks proposed for the World Robotic Sailing Cham pionship WRSC Namely these were a station keeping task an accuracy task and an endurance task A short description of the tasks proposed for WRSC 13 can be found in 47 Sailboat Position 15 4281 7 3 NAVIGATION TASKS 101 Tracking Visibl obstacle avdidance Unknovin obstacle afoidance Mobile obstacle avoidance Cooperation Battle Figure 7 6 WRSC arena 7 3 1 Accuracy Task In this task the sailboat has to follow autonomously a triangle see Figure 7 6 visiting its vertices in a precise order and trying to keep the course as close as possible to the triangle sides The objective here is to test how the sailboat sails under different points of sailing Figure 7 7 represents a test done on a triangle of the approximate dimen sions included in Table 7 4 Wind direction along the route is also depicted The route is colored differently for each side of the triangle That day the wind was quite unstable with frequent change of direction and intensity with some strong gusts Latitude 102 CHAPTER 7 RESULTS Accuracy 28 1325 Side Wind Up Wind Side Wind gt Wind Direction 28 1324 28 1324 28 1323 2
39. dged from the sailboat Next sections describe the commands that can be sent to the sailboat from the Base Station All the commands related with waypoints we will be detailed in the next section 6 1 7 Messages Sent To The Sailboat There are two different commands that can be sent to the sailboat au tonomous mode command or RC control mode command These commands are prepared in voidMainW indow sendCommBoat Autonomous mode command The format of autonomous mode command messages is W zxzrLxxMxxrAxrExrNxxCzxzx Not all fields need to be set on the message thus the format of this type of messages is flexible Basically it is used to fix a waypoint command the sailboat to transit into autonomous mode set some operational frequencies or combinations of these The meaning of the fields are defined in table 6 2 62 CHAPTER 6 CONTROL SYSTEM Table 6 2 Format of an autonomous mode message Wax Indicates that this is an autonomous mode command and the position of the way point on the list that will be set on this message Lex Indicates the longitude of a waypoint Maa Indicates the latitude of a waypoint Axe Indicates how far in meters can be the sailboat off the line that connects the cur rent position of the boat and the waypoint Ezxz Indicates the telemetry emission rate from the boat Naxx Indicates the execution rate of the naviga tion procedure Crx Indicates if the on bo
40. e boat is in remote control mode it obeys the sail and rudder po sition commands sent from the base station using a wireless game pad con nected to the laptop running the base station application In this mode short radio packets are sent to the vessel at a frequency of 5 Hz The trans mission rate can not be too high because in this mode the on board sensors are sampled logged and telemetry packets are sent at the specified frequency to the base station As we mentioned before on Section 6 1 7 control messages follow the format Craxsxx with xx denoting the commanded rudder and sail posi 82 CHAPTER 6 CONTROL SYSTEM tions The control program on board monitors the reception of these pack ages several times per control cycle and if received they are parsed and the servos commanded Autonomous Mode In the autonomous mode the navigation is fully under the control of the on board microcontroller The control is organized around three levels of control At the highest the route controller simply manages the list of waypoints that defines a route and selects the active waypoint When the sailboat is inside the radius of precision of the waypoint the route controller will change the active waypoint to the next one in the route The list of waypoints is treated as a cyclic route by default so when the last waypoint is reached it will start again with the first one and the route is repeated The list of waypoints h
41. ending on 2 2 SAILS 11 Wind l Direction In Irons Into the Wind Close Close Hauled i Hauled Close Close on I Reach Beam Reach Broad Reach Running Figure 2 2 Sails Positions wind direction wind velocity and heeling we should place the sails and the rudder in a way that we avoid that situation 2 2 Sails Sails are the most important part of a sailboat They provide the large potential autonomous sailboats have as high speed vehicles of virtually un limited autonomy for environmental monitoring and sampling Depending on their net displacement and dimensions they can accept scientific pay loads that maybe are too large or too power demanding to be integrated in other types of autonomous marine vehicles Al in all sails are the source of power of sail boats and they have many 12 CHAPTER 2 THE VESSEL HEAD y LUFF 1 Leech Roach Luff Allowance LEECH A Oo Ta A FOOT t Foot Round FOOT L Foot Round t Luff Edge eO Leech Edge ry ms a ie Camber Cross section Figure 2 3 Sail parts 28 parts as we can see in Figure 2 3 2 2 1 Sail Construction Sails come in two general varieties of construction single panel where the entire sail is one piece of cloth without seams gores or cuts in the body of the sail and paneled where each sail is made up of panels or strips of cloth attached edge to edge with tapered seams to induce three dim
42. ensional shape in the sail when filled with wind Single panel sails of a woven material are typically encountered in the 2 2 SAILS 13 construction of kit boats because single panel sails are much less costly for the kit maker to provide The details of successful setting trimming and tuning of single panel sails is an entire subject in itself Single panel sails must be full of air so that the cloth can stretch under the wind loading and can begin to take on the cambered shape necessary for drive Paneled sails are used in a greater proportion in racing classes and also in scale models that are going to be operated on the water They provide superior performance because the airfoil shape that produces drive for propelling the hull is built into the sail Most use a membrane material such as a mylar sandwich with load carrying fibers or mylar film These modern materials do not stretch so the cambered shape must be built into the sail with tapered seams that hold the panels together We use single panel sails because they are less complicate to trim than the paneled ones and in addition with only one configuration we can navigate in a variety of conditions 2 2 2 Sail Attachment A sail is attached to the spars by each of the three corners and by the luff of the sail to either the mast or the jibstay In the case of the jib luffs are fitted with a hem into which the jibstay slides This supports the sail along its length
43. es the active waypoint longitude Izx Indicates winch intensity Six Indicates sailboat speed as estimated by the GPS receiver Una Indicates speed calculated from GPS co ordinates Jrx Indicates distance calculated to the previ ous point Fax Indicates azimuth Krr Indicates GPS time Quzx Indicates milliseconds from the initializa tion are Indicates rudder position baa Indicates sails position Timing Measurements Message Timing measurements message contains the execution cost in milliseconds of selected parts of the on board control system The structure of this type of messages is as follows Dax Cost in milliseconds of the full cycle aLr Cost in milliseconds of the first check for incoming packets baa Cost in milliseconds of invoking checkTimeOut cre Cost in milliseconds of invoking the bear ing selection algorithm 90 CHAPTER 6 CONTROL SYSTEM dzz Cost in milliseconds of reading all sensors frx Cost in milliseconds of checking a second time for incoming packets grr Cost in milliseconds preparing a telemetry packet hax Cost in milliseconds of sending a message inx Cost in milliseconds of logging a data packet joz Cost in milliseconds of obtaining a new reading from the TCM2 50 board kzz Cost in milliseconds of obtaining a new GPS reading law Cost in milliseconds of some numerical computations miz Cost in milliseco
44. ess game pad Result The sailboat will change the mode Flow Push C on the wireless game pad Sets the station base application in the opposite mode than the actual mode Starts to send messages in the selected mode Table 6 14 Set sailboat mode 6 2 Sailboat Control System The on board system communicates with the base station through the ex change of messages Messages are parsed to detect if it is a remote control or an autonomous mode message If the system is in autonomous mode given the actual position the waypoint coordinates the actual heading and the instantaneous wind direction it selects the best bearing to reach the waypoint goal 6 2 1 Software Architecture The on board software has been developed modularly and all modules have been first developed and tested in isolation Because the development en vironment lacks any kind of debugging facilities we have designed a basic methodology to tackle this problem It is based of defining a set of compi lation directives to select which parts of the code are to be compiled and in this way tune the on board control code to different debugging condi 70 CHAPTER 6 CONTROL SYSTEM tions An additional set of compilation directives allows to select the level of verbosity of modules debugging outputs The next table describes all the compilation directives and how they configure the on board control program Obviously some possible conf
45. et sailboat Mode win sady ee a AA ee Se doe SD logging time improvements 1 e ee ee en Cycle time statistics roii inr 3 2 gas Nee a can A ee ede do vii viii 7 3 7 4 7 5 7 6 Al LIST OF TABLES Times 3 3 A werd A ao Se Bogs Goeth ed arto br a te 94 Accuracy task os O a SE o Se EE 102 Mesurements 1 ee e e 106 Mesurements e e g deon e e a a es 106 GPS improvement 117
46. hat has been addressed and solved is related to the roundup conversion error that is normally present when con verting longitude and latitude coordinates obtained from NMEA sentences to equivalent single precision coordinates in decimal degrees Note that the avr libc does not provide support for double precision real numbers Latitude and longitude coordinates are obtained from NMEA messages as strings with the format dddmm cccc where d stands for digits of de grees m for digits of minutes and c are the decimal part of minutes Parsing these strings implies converting the minutes to degrees and adding it to the degrees The operation is straightforward in principle but due to the limited capacity of single precision real numbers to represent real numbers up to six decimal digits it requires a more careful conversion procedure Other wise the roundup errors are normally below a meter but occasionally the A 4 AVR LIBC ERROR 117 combined error in latitude and longitude can reach 2 meters On Table A 1 we can see the difference between the previous function and the new one and the error in meters Table A 1 GPS improvement NMEA coordinate Decimal deg Before Decimal deg After Diff in meters 2830 2789 28 504646 28 504648 0 22m 12847 3562 128 789270 128 789276 0 6672m 8147 0002 81 783337 81 783333 0 4448m 12847 0001 128 783335 128 783340 0 556m 01547 0001 15 783335 15 783336 0 1112m 2806 7944 2
47. ick 3D hardware via OpenGL and 2D video framebuffer It is used by MPEG playback software emulators and many popular games SDL supports Linux Windows Windows CE BeOS MacOS Mac OS X FreeBSD NetBSD OpenBSD BSD OS Solaris IRIX and QNX The code contains support for AmigaOS Dreamcast Atari AIX OSF Tru64 RISC OS SymbianOS and OS 2 but these are not officially supported SDL is written in C but works with C natively and has bindings to several other languages including Ada D Eiffel Erlang Euphoria Go Guile Haskell Java Lisp Lua ML Oberon Component Pascal Objective C Pascal Perl PHP Pike Pliant Python Ruby Smalltalk and Tcl SDL is distributed under GNU LGPL version 2 This license allows you to use SDL freely in commercial programs as long as you link with the dynamic library In the sequel we detail the different stages of the Base Station operation 58 CHAPTER 6 CONTROL SYSTEM 6 1 3 Initialize The initialization of the system it is done by MainWindow MainW indow QWidget x parent This part of the application initializes the variables starts the timers to update the widgets and send the commands connects to the XBee module starts the background thread that controls the wireless game pad and at the end it loads the Google map corresponding to the area of interest Firstly the timers are configured to send data to the sailboat every 500 milliseconds and to update the widgets every 100 mi
48. igu rations are useless in practice as the control program demands all modules to be included GPS If defined GPS support will be included Otherwise the GPS receiver will not be used and the sailboat will be always on the same position Latitude 0 Longitude 0 TCM If defined the code related to the TCM2 50 sensor will be included Other wise the TCM board will not be operative and the sailboat will have always the same value of pitch roll compass and temper ature all of them 0 XBEE If defines the RF XBee module will be operative Otherwise the radio will not operative and the sailboat will not be able to communicate with the base station SD If defined support for operating the SD card will be included Otherwise the mi cro SD card will not be usable and logging will be canceled DEBUG SENSOR This directive permits debugging the code related with on board sensors Output is displayed on the serial monitor through the USB connection 6 2 SAILBOAT CONTROL SYSTEM DEBUG NAV This directive permits debugging the on board navigation algorithms Output is displayed on the serial monitor through the USB connection DEBUG XBEE This directive permits debugging the code related with radio communications Out put is displayed on the serial monitor through the USB connection DEBUG SD This directive permits debugging the code related with SD logging services Output
49. ilboat 29 This vessel was built in 1910 in the shipyards of San Telmo in Las Palmas de Gran Canaria Spain It won its first race on October 25 1911 One of its most memorable achievements took place in 1960 when under the command of skipper D Ventura Quevedo managed to circumnavigate the island of Gran Canaria in 27 hours and 8 minutes beating the previous record at the time In November 10t 2000 it was rebuilt and nowadays it is exposed at the building entrance of the Real Club Nautico of Gran Canaria A TIRMA is based on a carbon fiber One Meter Class vessel with main sail and jib displayed in Figure 2 4 It is equipped with two analog RC servos acting as actuators for rudder and sail sheets that are powered from a six NiMH rechargeable AA cell battery set with a capacity of approxi mately 20 Ah The boat has been equipped with a custom made wind vane situated on the top of the mast for sensing the apparent wind direction and other electronics that will be described later Figure 2 5 details the dimensions of the sailboat The most important 2 3 A TIRMA SAILBOAT 15 Figure 2 4 Our sailboat dimensions of the vessel are LOA 100 cm Beam 24 5 cm Draft 14 cm Sail Area 0 61 m Displacement 4 3 kg Mast Height 1 6m 16 CHAPTER 2 THE VESSEL 1 34m L 0 36m sE 0 14m 0 34m 0 07m ESA 0 04m 0 35m im Figure 2 5 Sailboat dimensions Chapter 3 State of Art This chapter is dedicated to
50. ion Figure 4 4 Bearing selection algorithm 30 4 2 AUTONOMOUS SAILBOAT NAVIGATION 37 polar diagram of the boat it significantly non linear and finally we need the true wind direction When the function has initialized all the variables it calculates the best bearing to the target by the right side and then by the left side To calculates de best direction the function calculates the projected velocity to the target If the selected direction gives a faster velocity to the target we update the new direction and then we test with other direction Now we have the best bearing by the right side and the left side The function will return the closer bearing to current direction to minimize the tack maneuvers 38 CHAPTER 4 NAVIGATION AND CONTROL Chapter 5 Onboard Electronics The vessel s electronics is made up of the following main components 1 An 8 bit microcontroller board 2 A XBee PRO 868MHz RF module 3 A GPS receiver 4 An electronic compass with inclinometers 5 A wind vane 6 A current sensor 5 1 System Hardware The sailboat controller hardware is based on a Waspmote V1 1 It is a commercial credit card size board based on a microcontroller ATmega1281 running at 8MHz The microcontroller integrates 8KB of SRAM for data 128 KB of FLASH program memory for program code and a 4 KB EEPROM 31 The board provides several UARTs an I C bus a micro SD card reader a real time clock a three axis accelerometer and
51. ions and tolerant to failure of on boards systems and sensors On the other hand they must be designed to be safe vessels that do not pose any threats to other boats This requires better perception capabilities and the incorporation of collision avoidance tactics It is important also to limit their speed and net displacement so they can be seen as potentially not harmful in case of failure Specific to the work that has been described in this master thesis future work might address some of the following lines of development The current version of the A TIRMA is quite limited due its physical dimensions We would like to apply the A TIRMA control system on a larger sailboat which might offer better navigation conditions and applicable on real field applications As commented before sailing a larger boat requires endowing it with improved sensing capabilities and security mechanisms to perceive the en vironment and avoid collisions A first step in that direction would be to develop a collision avoidance module based on an class B AIS receiver In the short term we foresee to continue this work with the development of field application of autonomous sailboats like optimal area sampling in 114 CHAPTER 8 CONCLUSIONS environmental monitoring applications Appendix A Improvements on Waspmote API The Waspmote s API used in this project has been version 0 32 for hardware version 1 1 However the API has been modified extensively i
52. ition 1 If Upwind And Heel Optimal Then Sail position 1 If Upwind And Heel Too High Then Sail position 2 If Close Reach And Heel Too Low Then Sail position 1 If Close Reach And Heel Optimal Then Sail position 2 If Close Reach And Heel Too High Then Sail position 3 If Through And Heel Too Low Then Sail position 2 If Through And Heel Optimal Then Sail position 3 If Through And Heel Too High Then Sail position 4 If Broad Reach And Heel Too Low Then Sail position 3 If Broad Reach And Heel Optimal Then Sail position 4 If Broad Reach And Heel Too High Then Sail position 5 If Run And Heel Too Low Then Sail position 4 If Run And Heel Optimal Then Sail position 5 If Run And Heel Too High Then Sail position 5 6 2 4 Control loop Once the initialization has been accomplished the main control loop will proceed in any of two possible modes of operation In the autonomous mode the sailboat s control system selects the best bearing to arrive to the active waypoint Alternatively the remote control or teleoperation mode permits to take full control of the sailboat from the base station In both modes the telemetry is kept active Figure 6 11 6 2 SAILBOAT CONTROL SYSTEM 81 Active task Route waypoints Mission controller State position bearing sensors Goal state goal position speed Figure 6 11 Control architecture Remote Control Mode While th
53. l Right Strong Right 0 8 0 6 0 4 0 2 200 150 100 50 0 50 100 150 Figure 6 5 Fuzzy input Desired heading 200 76 CHAPTER 6 CONTROL SYSTEM Input rudder position Tum Left Neutral Tum Right 0 8 0 6 0 4 0 2 0 30 20 10 0 10 20 30 Figure 6 6 Fuzzy Input Current rudder angle Output rudder Strong Left Left Neutral Right Strong Right 1 0 8 0 6 0 4 0 2 0 30 20 10 0 10 20 30 Figure 6 7 Fuzzy output Commanded rudder angle 6 2 SAILBOAT CONTROL SYSTEM The rules implemented were If Direction Strong Left And Turning Left Then Rudder Change Left If Direction Strong Left And Turning Neutral Then Rudder Change Strong Left If Direction Strong Left And Turning Right Then Rudder Change Strong Left If Direction Left And Turning Left Then Rudder Change Keep If Direction Left And Turning Neutral Then Rudder Change Left If Direction Left And Turning Right Then Rudder Change Strong Left If Direction Middle And Turning Left Then Rudder Change Right If Direction Middle And Turning Neutral Then Rudder Change Keep If Direction Middle And Turning Right Then Rudder Change Left If Direction Right And Turning Left Then Rudder Change Strong Right If Direction Right And Turning Neutral Then Rudder Change Right If Direction Right And Turning Right Then Rudder Change Keep If Direction Strong Right And Turning Left The
54. lliseconds Then the XBee radio is initialized and the function to connect to the XBee gateway is invoked indicating the USB port where the gateway is connected The control of the wireless game pad is carried out with a background thread using the QT class QFuture to start and check the wireless game pad At the end of the initialization phase we load the html file map html that controls the Google map and waypoints interface 6 1 4 Close The function MainWindow MainWindow stops the thread that con trols the wireless game pad and the XBee communications when the user closes the application 6 1 5 Update Widget The procedure voidMainWindow updateWidget every 100 millisec onds checks if we have received any data packet from the sailboat First it checks if we have received any packet from the sailboat during the last 15 seconds If that is not the case it changes in the interface to red the color of the connection state and changes also the control mode to autonomous because on board the sailboat changes automatically to 6 1 BASE STATION 59 autonomous mode and activates the Return To Home behavior in these circumstances If this is the case the application also updates the LED light that shows on the GUI if the vessel is in autonomous or in RC control mode When a packet is received the application displays the content of the message on the text browser and parses it to extract the data or interpre
55. m distance to the center 0 22 m Maximum distance to the center 12 76 m Mean distance to the center 3 48 m Standard deviation 2m Minimum square 15 2 x 17 24 m Latitude Latitude 7 3 NAVIGATION TASKS Station Keeping 28 1318 28 1318 28 1318 28 1318 28 1318 28 1318 28 1318 28 1318 28 1318 28 1318 107 Sailboat Position Waypoints Center Waypoints Center Position Boat gt Biggest Diagonal li 15 4284 28 1317 1 1 i 15 4284 15 4284 15 4284 15 4284 Longitude l l 15 4285 15 4285 Figure 7 11 Station keeping results Station Keeping 28 1324 28 1324 28 1323 28 1323 28 1322 Sailboat Position Center Waypoints Center Position Boat Biggest Diagonal 15 4262 i 15 4282 15 4282 15 4282 Longitude L 15 4282 Figure 7 12 Second station keeping results 15 4281 15 4281 Volt 108 CHAPTER 7 RESULTS 7 3 5 Endurance Tests Endurance tests are typical of WRSC For the 2013 edition the sailboat will have to demonstrate its capacity to sail for 8 hours while performing some sensor sampling tasks Our sailboat has two sets of batteries one powers the waspmote board and sensor meanwhile the second one powers the servos only We did done an endurance test over three and a half hours and got the discharge curves shown in Figures 7 13 and 7 14 Servo Battery 8 5 7 5 6 5 5 5 4 5 ati 0
56. mestamps of new readings must be checked against the timestamp of the last delivered message to verify that it is indeed a new reading If that is not the case a new reading is attempted The compass board is programmed to produce a continuous flow of mea surements at a default frequency of 10 Hz This approach reduces the cost of reading from the TCM board Each data packet contains the compass pitch and roll angles the temperature of the board and eventually an error code Commonly error codes appear when the magnetometers have become saturated or the pitch and roll angle limits have been exceeded In those cases these measurements are discarded TCM packets may accumulate and overflow the microcontroller serial buffer if it is not read fast enough This is not a problem because the serial port buffer is has been designed as a circular buffer At the same time when new measurements are read from the serial buffer all messages but the last one are discarded It is important to know that the GPS receiver and the compass are connected to two different serial ports that are in fact multiplexed on the same microcontroller s UART This implies that it is not possible to receive continuously and simultaneously data from both devices Our solution is to read first the TCM 2 50 sensors board and then flush the serial port and set the multiplexer to read from the GPS receiver The navigation is critically dependent on the adequate sampling rate
57. ming Measurements Message 89 7 Results 91 7 1 Code Improvements e 91 A A A OE 91 7 1 2 Multirate Control Cycle o o 95 IP QCHSOTS WESTS sew Vian dca Ge nl sh ae and od S 95 CONTENTS 7 2 1 Wind Load Sensor 2 00000 04 7 2 2 Wind Vane Noise 0 0008 eee Tota CEIM OO eda kita i te o ay ge ee a Ss 7 2 4 GPS Measurements 00 00004 7 3 Navigation Tasks 2 ee Wak Accuracy Task pcs ak y Bee eee ee 7 3 2 Return Lo Home RTH os a iho be ee es 7 3 3 Excessive Heeling Effects 7 3 4 Station Keeping Task 7 3 5 Endurance Tests 8 Conclusions 8 1 Future Work Lata a a do A E A Improvements on Waspmote API A XBee Support A A A Toi a ASP ierg AI a A A a a ts AA AVR Libe Error 2 Mass e e iii 95 97 99 99 100 101 103 104 106 108 111 113 CONTENTS List of Figures 2 1 2 2 2 3 2 4 2 5 3 1 3 2 3 3 3 4 3 5 3 6 3 7 3 8 3 9 3 10 4 1 4 2 4 3 4 4 5 1 5 2 5 3 5 4 5 9 5 6 5 7 5 8 5 9 5 10 5 11 Sails physics ia a da weed dy a a ta A ah ee 10 Sails Positions Lia Rs 2 he Bee Ya thaw a edo ad 11 Sail parts 28 4 5 6 450 bw Sew whe Ga Shaw ww woh eS dhe ee 12 Ourisailboat Am a arto a an ek a E et Eland ee we Ge cele 15 Sailboat dimensions 16 Wind Vane TIS oi ps ela AE ects el gee ee el N 18 Balanced rig concept 10 2 e a a 21 Protei
58. mits information to the sailboat by means of the XBee gateway Depending of the mode selected by the user the application transmits different information to the sailboat If the application is in RC control mode every 500 milliseconds it transmits to the sailboat the commanded position of rudder and sails but if it is in autonomous mode the base station resends the Home coordinates and some other parameters described on Table 6 2 every 2 5 seconds Depending on the working mode of the application namely RC control or autonomous mode the system goes through the following stages e Initialize e Close e Update widget e Edit the waypoints list e Send command to sailboat 6 1 BASE STATION 97 This application uses the libXBee and the Sample Directmedia Layer library to support respectively the radio communications and the wireless game pad libXBee v1 LibXBee 42 is a C C library that simplifies the usage of Digi XBee radios It provides a friendly interface for creating connections and sending and receiving data packets Eventhough this library is now in version 3 we have kept using version 1 because versions 2 and 3 were not backwards compatible It is planned to update the base station software to the most recent version of this library Sample Directmedia Layer SDL The Simple DirectMedia Layer 43 is a cross platform multimedia library designed to provide low level access to audio keyboard mouse joyst
59. n Rudder Change Strong Right If Direction Strong Right And Turning Neutral Then Rudder Change Strong Right If Direction Strong Right And Turning Right Then Rudder Change Right 78 CHAPTER 6 CONTROL SYSTEM Also to avoid problems with overheeling we check if the heeling of the vessel it higher than a maximum threshold If it is higher then the system apply a correction to the rudder to compensate the sailboat s tendency to haul upon the wind due to excessive heeling Sails Control The inputs for sail control are the desired point of sailing Figure 6 8 and the roll angle Figure 6 9 The desired point of sailing determined by the wind direction for the established bearing defines the position of the sail even though this position is adjusted depending on the roll angle to avoid excessive heeling As output we have five positions of the sail Figure 6 10 Input wind course Upwind Close Reach Through Broad Reach Run 0 8 0 6 0 4 0 2 Figure 6 8 Fuzzy input Point of sailing 6 2 SAILBOAT CONTROL SYSTEM 79 Input roll 1 0 8 0 6 0 4 0 2 0 0 10 20 30 40 50 60 Figure 6 9 Fuzzy input Roll angle Output sails Position 1 Position 2 Position 3 Position 4 Position 5 1 0 8 0 6 0 4 0 2 0 10 20 30 40 50 60 70 60 90 100 Figure 6 10 Fuzzy output Sail position 80 CHAPTER 6 CONTROL SYSTEM The rules implemented were If Upwind And Heel Too Low Then Sail pos
60. n a number of aspects A 1 XBee Support The XBee support in the Waspmote s API for hardware version 1 1 was substituted with and adapted version of the equivalent for hardware version 1 2 that has been modified in several aspects to improve reliability memory consumption and speed One aspect that has been addressed in depth is the OTAP Over The Air Programming support included in the Waspmote s API This feature was an important need in our context because it might facilitate enormously uploading new programs on board the sailboat during field trials In order to adapt it to our needs we implemented a new OTAP shell for Linux 115 116 APPENDIX A IMPROVEMENTS ON WASPMOTE API A 2 Timer In order to control the sails positions with an adequate resolution it was necessary to change the duty cycle of the PWM drive because with the the API s default duty cycle we could only discriminate among 6 positions We adapted the TimerThree Arduino library 51 that easily allows to set the period and the duty cycle of the any of the ATMega timers Now we have 20 different sail positions This number of positions makes the system to be more accurate than before A 3 GPS The GPS API module has been completely rewritten to achieve better con trol of the GPS receiver Currently the GPS receiver can operate at different UART speed supports a broader range of binary commands and is much more robust A subtle problem related to GPS t
61. nd position Xxx Indicates X accelerometer value Yzxz Indicates Y accelerometer value LLL Indicates Z accelerometer value WIX Indicates Waspmote battery value SIT Indicates servos battery value Nzxz Indicates the index of the actual waypoint Grr Indicates the desired bearing Vaz Indicates the temperature of the XBee ra dio module in Celsius degrees 6 1 6 Editing The List Of Waypoints The user has the possibility of performing some operations over the list of waypoints It is possible to e Add e Reorder e Delete e Resend To add a new waypoint the user has to click on a place in the map where he or she wants to add the new waypoint The map html file will register the point and when updateWidget asks for new waypoint it will return the new point that will be sent to the sailboat 6 1 BASE STATION 61 The user can also reorder the list of waypoints doing click on Up or Down buttons and the system will instruct the sailboat controller to reflect also the new arrangement of waypoints To delete a waypoint the user has to select it and click on delete and the system will send the command to the sailboat to delete the waypoint If we want to resend a waypoint we have to select it by clicking on it in the waypoints list and it will be resent to the sailboat This it is useful because if there were any connection problem it gives us the possibility to resend the waypoints that have not been acknowle
62. nds of reading the wind vane accelerometers and batteries nee Cost in milliseconds of reading the wind load current sensor As commented previously all collected sensor data are packed and logged on board on the micro SD but only a fraction of the logged packet is trans mitted to the ground station as a telemetry packet at a predefined fre quency Chapter 7 Results The data collected across several sea trials will be used in this chapter to evaluate the behavior of the sailboat under different circumstances It will also rise the opportunity to discuss the evolution of the control code and its timing performance We will base some of the tests on tasks proposed for the 2013 edition of the WRSC World Robotic Sailing Championship 47 7 1 Code Improvements 7 1 1 Timing As we described before on Section 6 2 7 during sea trials we have recorded the temporal cost of executing specific blocks of the code In a first stage that information was used as profiling data to improve the code and minimize the time spent in slow functions In Figure 7 1 we can see a non optimized time code It should be recalled here that our system is a 8 MHz 8 bits microprocessor with only 8 KB of RAM memory When we developed the code we did it knowing the capabilities of our system trying to do a fast code with low memory requirements It s worth commenting also that neither the development environment nor the avr gcc compiler offers
63. ngs of 22nd International HISWA Symposium on Yacht Design and Yacht Construction Amster dam The Netherlands 2012 BIBLIOGRAPHY 121 20 23 a 24 25 26 27 uu 28 Stelzer R Autonomous Sailboat Navigation Novel Algorithms and Experimental Demonstration PhD Thesis Centre for Computational Intelligence De Montfort University UK 2012 Allensworth T A short history of Sperry Marine 1999 http www sperrymarine northropgrumman com Company Information Corporate History Sperry History Ninorsky N Directional stability of automatic steered bodies Journal of American Society of Naval Engineers 34 2 280 309 1922 Stelzer R Pr ll T John R L Fuzzy Logic Control Sysrtem for Autonomous Sailboats IEEE 2007 Polkinghorne M Roberts G Burns R Winwood D The im plementation of fixed rulebase fuzzy logic to the control of small surface ships Control Engineering Practice 3 3 321 328 1995 http 202 114 89 60 resource pdf 2231 pdf Sauz C Neal N MOOP A Miniature Sailing Robot Platform Robotic Sailing Proceedings of the 4th International Robotic Sailing Conference Part II Pages 39 53 2011 Burnie M ed Participant Package World Robotic Sailing Champi onship 2010 and International Robotic Sailing Conference 2010 Queen s University Kingston 2010 Introduction to sailing concepts http en wikipedia org wiki Sailing Ca
64. o make detailed maps of the seafloor before start building a subsea infras tructure Also in the military sector AUVs are used routinely to perform acoustic vigilance or to localize mines Underwater gliders 4 are a relatively new technology that is revolu 1 1 UNMANNED MARINE VEHICLES 7 tionizing this field They move by gravity regulating their buoyancy but are normally equipped with wings that transform the vertical gravitational force into forward motion Due to this form of impulsion they move relatively slowly typically below 0 6 m s but their range autonomy is in the order of thousands of kilometers In practice this autonomy is limited nowadays by biofouling and the power demands of the scientific instruments carried on board 1 1 2 Surface vehicles The term ASV Autonomous Surface Vehicle refers to any vehicle that op erates on the surface of the water without a crew ASVs have been tested since World War IJ and numerous examples exist of military vehicles of this type but it has not been until recently that ASVs have started to be con sidered for scientific missions ASVs are valuable in oceanography as they are more capable than drift ing buoys but far cheaper than research vessels and more flexible than opportunity ship contributions 5 We can find three different types of ASV depending on the source of energy they use for impulsion T he widest class is that of motorized vehicles power by electric motor
65. oal of the project has been to transform a radio control sailboat into a complete autonomous system Through this document we will discuss its sensors and actuators describe the software on board and the implementation of a base station application to communicate with the sailboat Extensive tests have been carried out in real conditions concretely in Alcaravaneras beach in Las Palmas de Gran Canaria Spain Data collected during those sea trials will be used to describe her control system and discus her performance Survey speed varies with specific design and weather conditions The presented values represent the range of current autonomous sailboats in an approximate sea state of 3 or less Chapter 2 The vessel The One Meter Class sailboat is a developmental class which means that there are very few design restrictions The basic design restrictions include overall hull length keel depth and sail area The hull can be self designed or built from the scratch and easily planked from wood or any other suitable material The structure of the class allows the designer builder to experi ment with a design and then try it out on the water The design restrictions for the One Meter Class are the following 1 LOA Length Over All 99 100 cm 2 Maximum Mast Height 1 65 m 3 Maximum Mast Diameter 1 90 cm 4 Rudder must be aft of keel fin 5 Maximum Boom Diameters 1 90 cm The One Meter Class is light weight very fast and
66. of 6 2 SAILBOAT CONTROL SYSTEM 87 the set of on board sensors With the limited computing power available and high temporal cost of sampling some sensors a multirate smart sampling strategy is necessary Sensors like the wind vane and the compass have a high update rate and the reading cost is very small On the other side GPS has a low update rate and interrogating the GPS receiver takes about 60 msecs To deal with this situation fast sensors and in particular the wind vane are sampled several times within a single control cycle and filtered to produce better estimates of these magnitudes Sensors readings are monitored and they may trigger some alarms For example a sudden roll in wind direction over a predefined threshold will trigger the execution of the bearing selection algorithm during the next control loop If the vessel is turning we do not execute the bearing selection algorithm Also battery levels are checked against low level thresholds and if low battery alarms are triggered they are notified to the base station With the current hardware the temporal cost of executing one control cycle is dominated by the temporal cost of the actions carried out during one control cycle The subtasks that have the higher temporal cost are reading the GPS 60 milliseconds and SD logging 60 milliseconds Taking into account that some subtasks do not execute in every cycle the shortest cycle time takes approximately 160 milliseconds and
67. of the more expensive invocations 45 msecs Luckily it was found that it could be reduced to just a few milliseconds only changing from asynchronous to synchronous writing as shown in Table 7 1 Table 7 1 SD logging time improvements Before After Min 35 ms 0 ms Max 152 ms 144 ms Mean 45 33 ms 4 13 ms Standard deviation 6 52 ms 7 14 ms Figure 7 2 displays graphically the timing results after code optimiza tion Table 7 2 summarizes the effect of timing improvements after code optimization on cycle time Similarly Table 7 3 shows the effects over the main parts of the code Table 7 2 Cycle time statistics Before After Min 238 ms 140 ms Max 12409 ms 635 ms Mean 424 39 ms 285 86 ms Standard deviation 556 05 93 76 millizecon dz 94 CHAPTER 7 RESULTS Table 7 3 Times Function Min Max Mean Standard deviation checkForNewPackets 1 0 59 0 93 5 89 checkTimeOut 0 3 0 14 0 56 check Bearing 8 172 23 64 43 90 updateSensors 52 312 174 76 65 02 checkForNewPackets 2 0 60 3 53 13 95 prepareT elegram 60 121 65 93 10 51 radioSend 0 27 6 64 11 57 logT elegram 0 28 3 84 4 14 Times 350 300 250 200 150 0 20 40 60 60 100 120 140 160 180 cycles Figure 7 2 Timing with and without code optimizations 7 2 SENSORS TESTS 95 7 1 2 Multirate Control Cycle The A TIRMA sailboat does not exhibit fast dynamics that demand a con trol frequency much higher than the average 3 4 Hz discussed in the p
68. of two main parts a base station and a sailboat The base station is a laptop equipped with a XBee USB adapter board used to communicate with the microcontroller onboard the sailboat over a 868 MHz RF link On the sailbot an onboard system treats the received data and commands the sailboat to reach the waypoints Both systems communicate regularly at a predefined but modifiable fre quency Using this radio link the vessel can be monitored and controlled from the base station 6 1 Base Station The base station is a Linux application with a Qt 41 front end that relies on the libXBee 42 library to support the radio communications using XBee radio modules Through the graphical user interface GUI just by click ing on a Google map see Figure 6 1 it is possible to add edit or delete sequences of waypoints to define a route for the sailboat The interface displays the telemetry data received from the sailboat with information about sailboat s state parameters like bearing speed or posi tion along with sensor readings It is also possible to modify from the base 53 54 CHAPTER 6 CONTROL SYSTEM station some thresholds and parameters like the frequency at which the telemetry packets are remitted or the minimum frequency at which the bear ing selection function must be invoked This capability has demonstrated its usefulness during sea trials The application can switch between autonomous or RC control modes In this l
69. og sensors and communicate with sensors like GPS or compass over 5 1 SYSTEM HARDWARE 43 AUX SERIAL 1 TX re wet AUX SERIAL 1 RX DIGITAL DIGITAL DIGITALA4 DICITALS AUX SERIAL 2 RX DIGITAL DIGITAL2 AUX SERIAL 2 TX RESERVED DIGITALL RESERVED ANALOGS ANALOG GND ANALOG4 ANALOGS GND ANALOG2 ANALOG3 MUX_RX SENSOR POWER ANALOGA MUX_TX GPS POWER 5U SENSOR POLIER SENSOR POWER SDA SCL SCL SDA Figure 5 3 Waspmote pinout serial interfaces 5 1 4 Working Environment To develop software for the waspmotes we have used the Waspmote IDE 31 which is based on the open source IDE for the Arduino platform 32 following the same style of libraries and operation It includes all the API libraries necessary to compile the programs As we can see in Figure 5 4 the Waspmote IDE has a code editor an area for messages a terminal a toolbar with buttons for the main functions and a menu Program loading on the Waspmote is carried out using a serial connection over an USB port In Figure 5 5 we can see the toolbar of buttons where we have the most important functions of the IDE The next list describes the functionality associated to each button e Verify Compile Verifies the code and compile it e Stop Stops the compilation 44 CHAPTER 5 ONBOARD ELECTRONICS Barco_OTAP_32 2 new Commands Distance EEPROM Fue char p_buffer 30 A define F str strncpy_Ptp_ buffer PSTRistr sizeof p_ buffer p extern C int add_to_me
70. onic Compass 0 00000 ee ee 47 BAY GPE Receiver End a a a es 48 6 5 gt Wand Vanes ds para eee ded hs ea a a aS 50 5 6 Current Sensor 51 57 Power Demands ob eea fuk a a es Bw a 51 6 Control System 53 OL Base Stations ted is id dead tu beret Bee 53 6 1 1 Base Station Hardware 0 55 6 1 2 Software Architecture 56 libXB ewal uc ae a ai a a ek 57 Sample Directmedia Layer SDL 57 61 30 Initializers dia laa PR 58 GALA Clos a a Gp ea bee 58 6 1 5 Update Widget 58 6 1 6 Editing The List Of Waypoints 60 6 1 7 Messages Sent To The Sailboat 61 Autonomous mode command 61 RC Control command 0004 62 6 1 8 Use Cases s crad uan ee 63 6 2 Sailboat Control System o 69 6 2 1 Software Architecture 69 0 22 Initialization aia a a e ts 74 6 2 3 Fuzzy Logic System 74 Rudder Control o 75 Sails Control sai cs aa a ORS 78 6 2 4 Controlloop 4 62 4 668044 dai 80 Remote Control Mode 00 4 81 Autonomous Mode 82 6 2 5 Robust radio connectivity 84 Send 1 0 62 Ayana oe SVE dn he he en a Be el A 85 NN 85 6 2 6 Sensor Sampling 86 6 2 7 SD Garda cs ace dla A ee RE ee a eyes 87 Full Sensors Message e e 88 Ti
71. pe Class Microtransat Length 1 38 m Height 1 73 m Displacement 17 5 kg Hull Type Based on yacht model of type Robbe Atlantis Sensors and Computers The boat is equiped with a 800 MHz PC running Linux a GPS receiver a tilt compensated electronic compass and sensors for wind speed and wind direction Power Systems It is equipped with solar panels providing up to 285 Wp of power during conditions of full sun and a direct methanol fuel cell delivering 65 W as a backup energy source Communication Systems WLAN UMTS GPRS and an IRIDIUM satellite communication system Figure 3 9 Roboat I Austrian Society for Innovative Computer Sciences INNOC 3 4 AUTONOMOUS SAILBOAT COMPETITIONS 29 Table 3 6 VAIMOS Ecole Nationale Sup rieure de Techniques Avanc es Bretagne ENSA Team Ecole Nationale Sup rieure de Techniques Avanc es Bretagne Location Bretagne Berst cedex France Europe Class Microtransat Length 3 65 m Hull Type Based on yacht model of type Robbe Atlantis Sensors and Computers A Linux based embedded computer a weather station that measures the wind speed and direction as well as GPS position an AHRS Altitude and Heading Reference System a Iridium communication system and actuators for sail and rudder control Figure 3 10 VAIMOS Ecole Nationale Sup rieure de Techniques Avanc es Bretagne ENSA 30 CHAPTER 3 STATE OF ART 3 5 Field Applications Autonomous sailboats
72. ption Creates a new waypoint displays it on the map and sends it to the sailboat Trigger To do click on a place on the map Precondition The map has been loaded The XBee gateway has been connected Result The waypoint is displayed on the map The new waypoint is sent to sailboat along with its position on the list Flow Do click on the map Update map html and save the waypoint Wait to load the waypoint from map html to QT Create the message Send the message Table 6 4 Create waypoint Name Up waypoint Id UC2 Actor User Description Set the selected waypoint one position upper Trigger To do click on up Precondition One waypoint of the list has been selected The selected waypoint can not be the first one Result The waypoint will be set one position up and the upper waypoint one position down Flow Do click on the up The waypoint selected is set one position up The upper waypoint will set one position down The list will be updated Send the two modified waypoints Table 6 5 Up waypoint 6 1 BASE STATION Name Down way Id UC3 point Actor User Description Set the selected waypoint one position down Trigger To do click on up Precondition One waypoint of the list has been selected The selected waypoint can not be the las
73. responsive to the controls Two radio channels are required for control of sails and rudder The boat is easily disassembled and fits in a small size automobile for trans portation 10 CHAPTER 2 THE VESSEL Net force due to lift from sails and rudder Lift Force due to rudder WIND Rudder Figure 2 1 Sails physics 2 1 How Does A Sailboat Work The sailboats have been used since the Egyptians 5000 years ago 27 The sailing technology has changed a lot since then but the main idea remains the same The sails of a sailboat work like the wind of an aircraft Figure 2 1 The power to move the boat is produced by a pressure difference between both sides of the sails and this difference depends on the position of the sails in relation to the direction of the wind Figure 2 2 shows different positions of the sails with respect to the wind For example if the wind comes from the bow we have to place the sails in up wind position if we do not set the sail on that position we will lose all the power of the wind and the sails will flutter The sail positions might work also like a brake in order to make the sailboat reduce its velocity Fine sails regulation i e how they are sheeted strongly depends on wind intensity For example when sailing close hauled with strong winds placing the sails in an optimum position may provoke over heeling consequently reducing boat s velocity and even changing its bearing Thus dep
74. review the evolution of the technical develop ments in sailing The historical antecedents of robotic sailing date back to the automatization of the rudder control and the utilization of self steering mechanisms but continue nowadays with the recent appearance of kite based proposals to use the wind to propel high tonnage vessels with fuel cost reductions of up to 20 8 3 1 Self Steering Gear The first task to be automated was the rudder control 9 which is also re ferred to as autopilot Self steering gears can be divided on mechanical or electronic solutions 3 1 1 Mechanical Self Steering The first approach to a mechanical self steering system was a solution cre ated by fishermen who set the rudder of their boat to a fixed position 17 18 CHAPTER 3 STATE OF ART Then Herbert Hasler designed a more sophisticated mechanical solution the wind vane an example is shown in Figure 3 1 It basically consists of a wind vane which is connected to the rudder and when the angle of the apparent wind changes the wind vane detects the change and activates the steering device to return the boat to the selected course Figure 3 1 Wind Vane 18 3 1 2 Electronic Self Steering Electronic self steering controls the rudder position based on various sensors data This system needs a compass a wind direction sensor and additionally a GPS receiver The major contribution to the development of a self steering system was made
75. revious section It is basically determined by the cost of executing certain tasks and reading slow sensors as the GPS However we can improve the response of the system to external com mands for example when in remote control mode or the quality of data provided by some fast but noisy sensors if some fast operations are per formed several times per cycle This simple idea has been applied to duplicate the frequency at which we check the reception of new data packets on the radio and also to sample wind direction In the first case we get a much more responsive behavior of the A TIRMA to external remote control commands in the second case the wind direction filtered estimates are better With this approach we can double 8Hz the operational frequencies for some parts of the code 7 2 Sensors tests 7 2 1 Wind Load Sensor The A TIRMA lacks a wind speed sensor or anemometer On this work we have tested the usage of a current sensor an ACS712 board Section 5 6 to indirectly estimate the wind pressure or wind load on the sails from the current consumed at the winch servo while it tries to keep the sails in position under wind load This approach has to deal with some difficulties Firstly the intensity consumed at the winch servo is not constant as it s under PWM control This issue has been solved integrating the measurements provided by the ACS712 during 40 milliseconds the double of the PWM period Secondly the current cons
76. rr R Sails The Power of Source Model Yachting Magazine Win ter Issue 2006 Available on line at http www theamya org Intro_ Sails pdf 122 29 30 31 32 es 33 Pula 34 ra a 35 a Oo A 40 BIBLIOGRAPHY An article about the TIRMA yacht http escoben blogspot com es 2005 04 e1l html Stelzer R Pr ll T Autonumous sailboat navigation for short course racing Robotocs and Autonomous Systems 56 604 614 2008 Libelium s Waspmote manual http www libelium com vi1 files documentation waspmote waspmote technical _guide _eng pdf Arduino web site http playground arduino cc Libelium s OTAP manual http www libelium com downloads documentation ove _the _air _programming pdf XBee 868 Pro specifications http www digi com products wireless wired embedded solutions zigbee rf modules point multipoint rfmodules xbee pro 868 specs PNI s legacy TCM2 5 electronic compass manual http www tri m com products precisionnav files specs tcm25 _spec pdf MAX3232 Datasheet http www ti com 1it ds s11s410i s11s410i pdf NMEA 0183 Interface Standard National Marine Electronics Associ ation http www nmea org content nmea_standards nmea_0183_ v_410 asp A1084 GPS receiver hardware manual http ec mobile ru user _files File Tyco A1084 _HM _V1 0 pdf US Digital absolute encoder MA3 http www usdigital com products encoders absolute rotary
77. rship o on 0 10 0 10 Rudder Change pct of max FUZZY RULES FOR RUDDER FIS Rudder Change Fun Left Neutral Right Strong Strong Strong left aia lett lett S Stron S z 8 Y Left Keep Left left 7 Middle Right Keep Left 3 4 Strong s amp Right right Right Keep Strong Strong Strong right right right Riga Figure 4 1 Fuzzy rudder position 23 4 1 LOW LEVEL CONTROL 33 On the illustrations shown in Figure 4 1 shows the fuzzy sets for the different sails positions Where on the first fuzzy set we see the difference between the actual bearing and the desired bearing On the second we can see the angular velocity in degrees per second positive if the sailboat is turning to the right side starboard or clockwise and negative on the left side port of the sailboat or anticlockwise On the third one the different positions of the rudder appear Finally on the last illustration we have the fuzzy rules applied to set the rudder to a given position depending on the current bearing and the desired bearing and the turning rate For controlling the sails the authors propose to use a function 4 2 that given a wind direction and a wind speed returns the optimum heeling Finally the heeling angle if fixed adjusting sails sheeting Figure 4 3 max h max 0 h kof MMM gt max T Desired Heeling deg 180 135 90 45 0 45 90 135 180 Apparent Wind Direction
78. s or fuel engines Another recent and revolutionary type of vehicle is the wave glider 6 a vehicle impulsed by sea waves Finally there have been since ancient times vehicles that move impulsed by the wind that is sailboats The main advantage of these two lasta types of ASV have over motorized ASVs is their virtually unlimited range autonomy Wave gliders main drawbacks is their limited payload capacity and relatively low speed of 1 m s Sailboats offer good payload capacity and depending on sail dimensions 8 CHAPTER 1 INTRODUCTION Table 1 1 Oceanographic platform relative capabilities in normal operating conditions taken from 16 Vehicles Manned Ship Glider Drifter Sailboat Platform Specs Survey speed m s 0 8 0 0 3 0 0 3 0 2 5 1 Survey time scale Weeks Months Months Months Cost USD 10 000 day 60 000 5 000 50 000 Adaptive sampling Yes Yes No Yes Max 24h range km 600 25 25 200 Payload vol weight 100m 500kg 0 1m 5kg 1m 25kg 1m 50kg are capable of larger speeds The main disadvantage is their limited capacity to surpass storm conditions unattended at least with conventional sailboat designs Table 1 1 taken from 16 summarizes the previous discussion on several types of marine vehicles comparing them across a number of characteristics 1 2 Outline of This Thesis This master thesis is committed with the design and implementation of an autonomous low cost and low power sailboat The main g
79. sailboat off the line that connects the current position of the boat and the waypoint it is equivalent to the PC parameter in 30 finally E indicates the emission period for telemetry messages When the vessel is in this mode the base station sends every 5 seconds short messages to verify the radio link If these packets are not received at the vessel for 20 seconds the active waypoint is deactivated and substituted by the coordinates that identify the Home Point and the sailboat will try to arrive to that point autonomously This constitutes the Return To Home or RTH behavior that has proved a valuable capability during field tests 6 2 5 Robust radio connectivity Loss of radio connectivity is something that may happen during sailing due to a variety of reasons and it is important to endow the sailboat with some recovery and continuity strategies to deal with these situations In order to 6 2 SAILBOAT CONTROL SYSTEM 85 increase the robustness of radio communications on this uncertain scenario the XBee radios are used in API mode and all exchanged messages have been limited in extension to make them fit within the payload of XBee API frames Basically this constraint reduces the complexity of recovering partially lost packets as all messages involve a single radio frame The payload size is 100 bytes so we have to adjust the data transmitted in packages to not exceed this limit Accordingly at the lowest level the XBee
80. shaft ma3 ACS712 product page https www sparkfun com products 8883 BIBLIOGRAPHY 123 41 42 43 44 45 46 47 48 49 50 Qt UI framework http qt project org LibXBee library http code google com p libxbee SDL library web site http www libsdl org J A Spaans Windship routeing Journal of Windship Engineering and Industrial Aerodynamics 19 1985 215 250 EFLL fuzzy logic library https github com zerokol eFLL Web site of the World Robotic Sailing Championship 2012 Cardiff Wales September 2012 http www microtransat org wrsc2012 Rules for the 2013 edition of the World Robotic Sailing Championship http www ensta bretagne eu wrsc13 pdf Rules1_2013 07 17 pdf Koch M Petersen W Using ARM7 and C OS II to Control an Autonomous Sailboat Robotic Sailing 2011 pp 101 112 Springer 2012 Neal M Sauze C Thomas B and Alves J C Technologies for Autonomous Sailing Wings and Wind Sensors in proceedings of the 2nd IRSC Matosinhos Portugal July 6th 12th 2009 pages 23 30 2009 Alvira M Barton T Small and Inexpensive Single Board Computer for Autonomous Sailboat Control Robotic Sailing 2012 pp 105 116 Springer 2013 Arduino Timer library http playground arduino cc Code Timeri AVR Libc Documentation http www nongnu org avr libc user manual index html 124 BIBLIOGRAPHY 53 AVR Libc Solution by Joerg Wunsch http code google
81. ssage message_t message const char format 7 char data l90 10 E void setup 1 eeprom _write_byte unsigned char 0x0l 0x00 ifdef USB ON USB begin USB print Fi Free MEMORY USB print freeMemory DEC USB print Fi bytesin endif ifdef _XBEE ON x meds ere saat a gt Figure 5 4 Waspmote IDE Compile New sketch Save Serial Monitor Stop compiling Open Upload Figure 5 5 Waspmote IDE Buttons 5 1 SYSTEM HARDWARE 45 e New Creates a new sketch e Open Opens a menu to find a sketch and open it e Save Saves the changes of current sketch we are editing e Upload to Waspmote board Compiles the code and upload it to the Waspmote board e Serial Monitor Starts the Waspmote s serial console 5 1 5 Over The Air Programming OTAP The concept of Over The Air Programming 33 or OTAP is commonly used in the scope of sensor networks for denoting the possibility of programming a sensor node wirelessly In this project this capability has alleviated the hassle of having to open the deck to access to the inner of the vessel and to extract the plastic container that hosts the Waspmote just to connect a USB cable every time a new firmware had to be uploaded to the sailboat The OTAP process involves the following steps e Locate the node to upgrade e Check current software version e Send the new program e Store the new program on the SD card e Reboo
82. sumption and operative endurance Abstract Chapter 1 Introduction This master thesis is concerned with the development of unmanned vehicles suitable for performing environmental monitoring and data sampling at sea The importance of the oceans in the overall health our planet Earth can t be underestimated For a long time seas were considered vast spaces whose once thought inexhaustible resources could be blindly exploited for fishing hunting mining or even used for dangerous wastes disposal Luckily human kind is now conscious about the importance that the health of the seas has for the whole planet and its population The international scientific community has expressed clearly an repeat edly the necessity of improving the knowledge we have about the oceans Specifically this knowledge is crucial to address the problem of global cli mate change as evidences continue to accumulate in that respect Three are the key problem associated with in situ data gathering at seas The first two are related to spatial and temporal data resolution The other one is cost Until very recently data could only be obtained from localized points using moorings and buoys with very poor spatial resolution Research and opportunity vessels have been the other classical mean for sampling at sea with good spatial and temporal resolution However the very high cost of a research vessel normally limits the duration of campaigns and impedes 6 CHAPTE
83. t the message Received messages can be GPS connection message a summary of the state of the sensors onboard or a confirmation reply to a new waypoint previously sent Telemetry messages are sent from the sailboat routinely They contain a summary of on board s sensors readings and follow the following format dixCxxPxxRixTxxWxxXxxY xxZxixwxxrsxiNxxGxxVxzx In Table 6 1 the meaning of the fields of the formate of telemetry packets is clarified To display the actual position of the sailboat we call the showNewMarker function which along with the evaluate JavaScript function allows us to send javascript commands from QT to a javascript code in a html file Finally the application checks if the user has introduced a new waypoint in the map and in that case the new waypoint s coordinates are sent to the sailboat using sendXbee Every time waypoints are updated the Base Station recalculates the distances between consecutive waypoints and updates the GUI with the new distances 60 CHAPTER 6 CONTROL SYSTEM Table 6 1 Format of telemetry messages d D aa If it is in capital letters it means that the boat its in autonomous mode and xz in dicates the duration of the previous cycle in milliseconds Cra Indicates the compass value Pax Indicates the pitch value Raa Indicates the roll value Teg Indicates the temperature value Wax Indicates the apparent wi
84. t and start with the new program e Restore the previous program if the process fails More details on the OTAP process can be found in Libelium s OTAP manual 33 46 CHAPTER 5 ONBOARD ELECTRONICS 5 2 RF Radio Module We have based all communications with the sailboat on XBee 868 Pro RF modules Figure 5 7 These modules operate at the 868 MHz ISM band using only one channel The bandwidth is 24 Kbps and the communications can be encrypted The nominal range using a 4 5 dB dipolar antena in a free field is 40 km but more realistic estimations are in the range of 10 km It is possible to adapt the transmission power in five levels from 1 mW till a maximum of 315 mW It works at 3 3 V and its current consumption is 500 mA in transmission and 65 mA in reception 34 Figure 5 6 XBee Module The position of the antenna on the sailboat is very important to have a good communication with the station base We first placed it on the mast but with that configuration the aluminum mast blocked the reception in some courses impeding the communication with the sailboat After some testing we found that the antenna could be fixed to the backstay without disturbing the wind vane using a deformable support with good results 5 3 ELECTRONIC COMPASS 47 Figure 5 7 The 868 MHz 0dBi antenna in its final placement 5 3 Electronic Compass The electronic compass is a legacy TCM2 50 board 35 Basically it pro vides tilt compensated
85. t navigation for short course racing 30 These papers have been reviewed in the recent PhD thesis of Roland Stelzner 20 The first paper Fuzzy logic control system for autonomous sailboats 23 describe how to transform the sailors knowledge into a Mamdani type fuzzy system to control a sailboat with two actuators rudder and sails It pro vides the low level sailboat controller The second paper Autonomous sailboat navigation for short course racing 30 proposes a method to select the optimal bearing to reach a destination given the true wind direction and the current sailboat position and heading This will provide a second control level for the sailboat 31 32 CHAPTER 4 NAVIGATION AND CONTROL 4 1 Low Level Control The authors of this article 23 propose an algorithm to control the rudder and sails of a sailboat The paper describes a fuzzy logic system to manage the control surfaces of the vessel sails and rudder like a sailor would do It is described a fuzzy method to control the rudder that takes into account the difference between the sailboat bearing and the desired bearing and also the angular velocity of the sailboat to avoid oversteering 4 Strong left Left Middle Right Strong right Degree of Membership o a 0 Desired direction dea Left Neutral Right a 5 2 5 05 5 o 2 a 0 50 0 50 turn deg s Strong left Left Keep Right Strong right 1 Degree of Membe
86. t one Result The waypoint will be set one position down and the bottom waypoint one position up Flow Do click on the up The waypoint selected it set one position up The bottom waypoint will set one position down The list will be updated Send the two modified waypoints Table 6 6 Down waypoint Name Delete way Id UC4 point Actor User Description Delete the selected waypoint Trigger To do click on delete Precondition One waypoint of the list has been selected Result The waypoint will be deleted Flow Do click on delete The selected waypoint is deleted The list will be updated Send to delete the waypoint selected Table 6 7 Delete waypoint 65 66 CHAPTER 6 CONTROL SYSTEM Name Send au Id UC5 tonomous message Actor User Application Description Send an autonomous mode message to the sailboat It will be sent if the user clicks on a waypoint on the list or every 2 5 seconds Trigger Every 2 5 seconds Do click on a waypoint Precondition To be in autonomous mode The waypoint have to exist Result The waypoint will be sent to the sailboat and the vessel will answer with a confirmation of reception Flow Do click on the waypoint or reach the timeout of 2 5 seconds Configure the message format WxxLxxMxxArxExxN
87. t the direction of the apparent wind but not its speed It works at 5 V and consumes 16 mA Figure 5 10 MA3 absolute angular encoder Figure 5 11 Home made wind vane 5 6 CURRENT SENSOR 51 5 6 Current Sensor The current consumption at the actuator that control the sails sheets is measured by means of an ACS712 board Figure 5 12 40 The instanta neous current consumption is read as a voltage at a microcontroller s analog input The ACS712 board integrates two potentiometers to adjust the in tensity range being sensed and the acceptable levels of output voltage This reading is used as an indirect measure of wind pressure in the sails It is powered from 5 V and consumes 7 mA Figure 5 12 ACS712 5 7 Power Demands A summary of power demands of the main components of the system along with the capacity of both batteries is detailed in Table 5 2 52 CHAPTER 5 ONBOARD ELECTRONICS Table 5 2 Power demands of system components Component Volt V Current mA Power mW Microcontroller 3 3 9 29 7 GPS 3 3 26 85 8 XBee 868 PRO 3 3 65 500 330 TCM2 50 5 20 100 MA3 encoder 5 16 80 ACS712 board 5 T 35 Electronics Total 660 5 Electronics battery 3 7V 6000mAh 22200 mWh Component Volt V Current mA Power mW Rudder servo 5 10 500 100 Sail winch 5 10 800 500 Actuators Total 600 Actuators battery 7 4V 2700 mAh 19980 mWh Chapter 6 Control System The system consists
88. the largest 500 milliseconds 6 2 7 SD Card The micro SD card is used to log all data collected and generated by the sailboat This information give us the opportunity to record the state of the vessel on each control iteration The system store in the SD card two types of messages A full sensor message and a timing measurements message The SD card specifications e FATI6 88 CHAPTER 6 CONTROL SYSTEM e Allocation table size of 16 bits e 4GB e HC cards are compatible Full Sensors Message The format of this message it is an extension of the one detailed on Table 6 1 The structure of a telemetry or sensor data message is d D aax If it is in capital letters it means that the boat its in autonomous mode and xx in dicates the duration in milliseconds of the last control cycle Cxzx Indicates the compass value Pax Indicates the pitch value Rez Indicates the roll value Tze Indicates the temperature value Wax Indicates the apparent wind position Xx Indicates X accelerometer value Yzxzx Indicates Y accelerometer value Liz Indicates Z accelerometer value wre Indicates Waspmote battery voltage SIT Indicates servos battery voltage Nzxzx Indicates the index of the active waypoint Gua Indicates the desired bearing Vax Indicates XBee temperature value Hzr Indicates the active waypoint latitude 6 2 SAILBOAT CONTROL SYSTEM 89 Ax Indicat
89. ton mini USB Solar Socket Leds Switch USB Power Led ON OFF Figure 5 1 Waspmote board V1 1 Top side Battery Aux GPS Sockets so ds m RTC e BTN 23230 SD Card Figure 5 2 Waspmote board V1 1 Bottom side 42 CHAPTER 5 ONBOARD ELECTRONICS 5 1 2 Electrical Characteristics The electrical characteristics of the Waspmote board are detailed in the following list Operational values Minimum operational battery voltage 3 3V Maximum operational battery voltage 4 2V USB charging voltage 5V Solar panel charging voltage 6 12V Battery charging current from USB 100 mA max 100mA Battery charging current from solar panel 280mA Button battery voltage 3V Absolute maximum values Voltage in any pin 0 5 V 3 8 V Maximum current from any digital I O pin 40mA USB power voltage TV Solar panel power voltage 18V Charged battery voltage 4 2V As we can see we must to be careful with all the modules that we connect to the Waspmote because we should not exceed the maximum current and voltage 5 1 3 Input Output A Waspmote may communicate with other external devices that it is possi ble because the board has many input output pins as we can see in Figure 5 3 The board is provided with digital input output pins analog inputs PWM and power outputs SPI 12C and UART interfaces etc This is one of the most important features for us because we have to control servos read anal
90. umed is critically dependent on the state of motion of the Poll Degrees Winch Current Value Sail 96 CHAPTER 7 RESULTS servo Only if the servo is at rest if it has arrived to the commanded position the current consumed will be used to maintain the servo in position under the wind load In the current implementation the wind load measurement is only produced when the sails position has been maintained along the last two cycles Figure 7 3 depicts a typical situation note that the highest the current consumed at the winch the lowest the reading returned by current sensor First the winch is commanded to a new position denoted by the red line Once the servo is in position and using the roll angle as a secondary evi dence of wind strength it can be appreciated how the ACS712 measurement follows approximately the evolution of the roll angle Wind Load Winch Current Sail position Roll cycles Figure 7 3 Wind load estimation 7 2 SENSORS TESTS 97 7 2 2 Wind Vane Noise In the A TIRMA control system it is of paramount importance to have available a stable estimate of wind direction to adjust the sails position and to select the best bearing to reach a waypoint However the wind vane is in essence a noisy sensor due to its principle of operation and also due to it is placement on top of the mast that makes it quite sensitive to pitch and roll movements of the sailboat and sudden wind gusts To cope
91. with these circumstances some form of filtering of the raw data provided by the wind vane is unavoidable In the A TIRMA a median filter with window of length 7 has been used with good results The median filter was selected by its well known capability to eliminate impulsive noise Figure 7 4 shows the raw readings obtained from the wind vane and the resulting signal after applying the median filter The outcome is a more stable wind direction estimate that translates in a better control of the sailboat 98 CHAPTER 7 RESULTS Figure 7 4 Wind Filter 7 2 SENSORS TESTS 99 7 2 3 TCM2 50 The TCM2 50 board provides compass pitch roll and temperature mea surements Although minimized by the placement of this board close to the base of the mast these measurements also suffer from unintended move ments produced during sailing The TCM2 50 can filter its sensors reading and provide cleaner signals The filtering parameters can be set by the on board control system This possibility has been exploited to obtain cleaner measurements of bearing pitch and roll angles directly from the TCM board The TCM2 50 board can be configured to output measurements only when interrogated or continuously In this case it was configured to output continuously the readings of all sensors except the 3 axis magnetometer readings at 8 Hz In that way the time required to obtain measurements is reduced considerably making the whole system more agile 7 2
92. xplored and described in the literature but a truly robust design is still to be achieved An alternative design for a wind sensor direction and intensity has been described in 50 Finally as a result of the work carried out in this master thesis a publi cation have been accepted to the IRSC 2013 e J Cabrera G mez A Ramos de Miguel A C Dom nguez Brito J D Hern ndez Sosa J Isern Gonz lez and E Fern ndez Perdomo e An Embedded Low Power Control System for Autonomous Sailboats e 6 IRSC International Robotic Sailing Conference e 2 to 6 September Brest France Europe And also that A TIRMA is going to take part in the contests organized during the IRSC 2013 8 1 FUTURE WORK 113 8 1 Future Work This area of field robotics is still in its childhood and we think there are many opportunities to contribute to its development Autonomous sailboats have many appealing characteristics to make them optimal vehicles for marine sampling and monitoring applications First of all this kind of vehicles allows simultaneous sampling of atmospheric and sea variables They may offer a virtually unlimited range autonomy at relatively high speeds and ample payload capacity However autonomous sailboats still have to evolve in many aspects to become more robust and reliable vehicles Here robustness implies multiple dimensions On one side it demands that the sailboat must be seaworthy capable of standing strong sea condit
93. xxCzrzx Send the waypoint If message is confirmed set the waypoint as received Table 6 8 Send autonomous message Name Send control Id UC6 message Actor Application Description Send to the sailboat a control message It will be sent every 0 5 seconds Trigger Every 0 5 seconds Precondition To be in control mode Result The control message will be sent to the sailboat Flow Reach the timeout of 0 5 seconds Set the package format Crarsxx Send the command Table 6 9 Send control message 6 1 BASE STATION Name Receive sen Id UC7 sors message Actor Application Description Receives a message from the sailboat parses the message and display the content Trigger Receive something from the sailboat The message must begin with D or d Result The message will be parsed and GUI will be updated Flow Receive a message Detect if the message begins with D or d Parse the message drxCxxPxrRxxTrrWxrXxrY xx ZxarwxrxrsrtiNaiGiaVxx Display the information Table 6 10 Receive sensors message Name Calibrate sen Id UC8 sors Actor User Description To calibrate the wind vane pitch and roll sensors Trigger Check the calibration option Result Sensors will be calibrated Flow To set the calibration
94. y 10 discrete positions on their system MOOP University of Aberystwyth UK 25 to avoid permanent switching between adjacent positions to save 20 CHAPTER 3 STATE OF ART power on the sail actuators 26 There is another method described by Stelzer et al 23 which does not calculate directly the sail positions using wind data It first calculates the desired heeling of the sailboat and then a feedback loop implemented as a Mamdani type fuzzy inference system controls the sail positions to reach the desired heeling of the boat 3 3 Recent Technical Innovations Sailing is an ancillary art but still in constant evolution Every year interna tional sailing races are used to announce new sometimes radical technical innovations based on new designs exotic materials or CFD results all them aimed at achieving faster vessels Autonomous sailboats may of course benefit from these technological advances but the goals that drive the development in this field of robotics are completely different Here much more important than speed are issues like robustness seaworthiness and energetic efficiency In this section we will briefly review some ideas and technological innovations that are nowadays of interest for autonomous sailboats Rig innovations Rigid wind sails and balanced rigs 10 are two of the innovations that have been experimented recently on autonomous sailboats Wind sails are rigid sails that have been used in some

Download Pdf Manuals

Related Search

Related Contents

  journée EMS2012 dpt91 - Agence Régionale de Santé  取扱説明書（PDF）はこちら  SD 22 - SAGI Spa    Lettre Mai 2011 - Préfecture de Mayotte  USER GUIDE - CAMBOARD Electronics  DreamLine DL-6717-01CL Installation Guide  Submersible Bilge Pump  Grindmaster AMERICAN METAL WARE 87710 User's Manual  

Copyright © All rights reserved. Failed to retrieve file