.ation Too for a Road- departure Avoidance System
Contents
1. t lation Tool for a Road departure Avoidance System Imu E User s Manual CAPC v4 2 Pw eh ARA ne S Or aO ies 8 2 MP 2 a s orate 227 2 tut E EE Tere re tete Een T 272A 22 e E EEUU A MERE ee 7 4 4 FN x M W wo x w B RR M E n ee E TEE Ur rer ur E MEE E er MAR ee ett ME E ERES an MN Coon i lt N M 0 Duo MM S erm Um SIM A ae en ONT z ERS 2 e EXAM a os m Re e a m wr an ee NES S a E at 5 A phan a x E Pa er B e x pre erae reU x urnam n RE 27A rn ua eter ne Le 22 5 sited Dico eee SORER Ra QURE NR ORARE ate 1 ee MU Up 25075 e Males s e dy Y re uU pees 25225 RK N TRAC s ee gt T 2 m o TN XD later Sim see eee x t __ SONS Mos AUR bY CARE ORE ER ea a CrP a OTC CEES s st eren 5 ES DOS PM DUI a2. 8 Drivin died inan e des md 9 Post processing test data mode suse pal d dt E 9 THe MM aa 10 T TT TTE TENES 10 The Models Mie Mi eo odi _ NE The Simulation iae 18 The Disturbance Menu sesesssesssssscresesesssscscesssosssssasesoessoene 19 The Roadway Menu orem Ee Oro oed e eI eee nna 21 The Controllers 22 The Scenarios 23 The Window 24 sect KD DADO NN a Ub iM 24 About this Manual This user manual introduces the use of the CAPC Crewman s Associate for Path Control simulation tool v4 2 This software is for the design and evaluation of road departure prevention systems and was used to design and eventually implement the CAPC road departure prevention system on a vehicle at the University of Michigan The software features a comprehensive set of user friendly pull down menus to allow selection and revision of countless dynamics model modules sensor systems and warning and control parameters This guide introduces the basic capabilities and elements of using the system Other references are provided at this report s end for more detailed information about the dynamic models and the control and logic modules as well as the CAPC road departure Work Files needed To begin using the
3. 6009 Faulted Rough fisphalt Wavy Surface Treatment FORD MPG north straight FORO MPG south straight Artificially generated Slabs This menu offers the following selectable items e No Unevenness The road surface is smooth e Good PCC where PCC stands for Portland Cement Concrete e Faulted PCC e Rough Asphalt e Wavy Surface Treatment e Test track north straight This is a road unevenness data file derived from a measured data file recorded at a nearby test track on the north straightaway e Test track south straight This 1 a road unevenness data file derived from a measured data file recorded at a nearby test track on the south straightaway e Artificially Generated Slabs CAPC v4 2 User s Manual 19 The length of slabs is 9 m They are modeled as smooth flat plates with a slight positive slope The joints between the plates is discontinuous and the discontinuity is modeled as a random process The height difference during the transition from plate to plate ranges between 6 4 and 12 7 mm Wind Roadway Controllers Scenarios Window Road Irregularities Wind id Wind Continuous Wind Crosswind Gust Random Crosswind DC Random Crosswind Modify Wind Parameters The wind disturbance menu offers the following items to be selected e No Wind The vehicle will not be exposed to wind disturbances There will be no aerodynamic forces acting on the ve
4. The simple driver model is characterized by a preview model that looks at a single point in front of the vehicle The other option is a more elaborate optimal preview model that minimizes the tracking error at several points in front of the vehicle The CAPC vehicle was designed to use two vision cameras one for the near field range and one for the far field range In the CAPC simulation tool the images of the cameras are determined by applying several coordinate transformations to the known roadway geometry in front of the vehicle assuming a flat earth model The hardware specifications camera CCD chip size and resolution have been modeled Furthermore the position of the cameras on the vehicle the orientation and the update frequency can be chosen arbitrarily The following subsystems are available in the CAPC simulation tool Vehicle Models CAPC v4 2 User s Manual 5 e 7 DOF flat vehicle model longitudinal lateral yaw and 4 wheel rotational DOF e 8 DOF yaw roll model 7 DOF model 4 roll DOF e 4 DOF full vehicle model 6 DOFs for the vehicle body 1 DOF for each axle and 4 wheel rotational DOFs Tire Models e Steady state different road friction coefficients possible Cornering stiffness as a function of vertical tire load Magic Formula pure slip combined slip e Transient first order relaxation system tire load dependent Anti Lock Brake System Model e First order lag system with brake pressure satur
5. This is the fastest integration routine available The integration of the differential equations is less accurate than with the more elaborate methods The integration time step is constant e Runge Kutta 2nd order This method is a scaled down version of the Runge Kutta 4th order integration method It is faster than 4th order routine but less accurate The integration time step is constant e Runge Kutta 4th order This is the most elaborate integration method available in the CAPC software The integration time step 1 constant e Start CAPC v4 2 User s Manual 18 Selecting this item will start the simulation A dialogue box appears allowing the user to assign a name to the ERD output file The simulation can be interrupted by pressing a button on the mouse e Continue After termination of a simulation run by pressing a mouse button it is possible to continue with the simulation from the current position The initial values correspond with the values after the termination of the previous simulation run e Reset This option resets all the state variables in the simulation All initial conditions will be set to zero with exception of the vehicle speed The Disturbance Menu The vehicle can be subjected to different kinds of disturbances Two sources are available road unevenness and wind Road Irregularities WE Controllers Scenarios Window Road Irregularities d No Smooth fisphalt
6. prevention system A menu driven user interface allows the user to define dynamics models for the vehicle including tires suspension disturbance effects roadway effects the roadway geometry and the road departure prevention system which can contain a warning and or an active intervention mode CAPC v4 2 can berunin three modes numerical simulation desktop driving simulator and post processing for comparing actual vehicle data with simulation Version 4 2 is available for both the PowerPC generation of Macintoshes and the previous 68K Macintoshes with FPU 17 Key Words 18 Distribution Statement automotive safety simulation This document is available to the public through driving simulator road departure National Technical Information Service vehicle dynamics 5285 Port Royal Road Springfield VA 22161 19 Security Classif of this report Unclassified 20 Security Classif of this page Unclassified 21 No of Pages 22 Price 30 Table of Contents About this Manual 2 Lot more I0iormali Tm 3 D schpsomof the CAPC 2 Tool 4 doe EO ea E 4 Dynamics Models X E 5 How to Use the Simulation 7 The Three Modes TCR HUNE 7 Simulation
7. PPC or an older Motorola 680x0 68K processor The main objective of the tool is to integrate subsystems into one lane departure avoidance system and to study the performance and interaction among the different modules as weil as to evaluate the overall system behavior and performance The seven major parts of the CAPC simulation tool are the vehicle the lane marker recording and processing the estimation of the future trajectory of the vehicle the time to lane crossing calculation the brake steer controller the driver status assessment and the CAPC supervisory controller The tool has three modes 1 simulation mode 2 desktop driving simulator mode and 3 combining actual road test data with simulations The modes have different uses but all use the same models for dynamics control and sensing and all three modes have a similar look and feel The tool was used for design evaluation and preliminary testing of the CAPC system An example of the use of the tool was an investigation into the influence of external disturbances road unevenness and wind on the estimation of roadway geometry vision system is mounted rigidly onto the body sprung mass of the vehicle so estimates _ of the lane edge locations are affected by unknown roll and pitch motions of the sprung mass The simulation tool led us to add pitch and roll sensors but to realize that only the lower frequency motions need be known simplifying both hardware and s
8. Roadway Controllers Lane Sensors gt Vehicle gt Tires Anti Lock Brake System Driver Path Prediction Cruise Control Driver Status Modify Parameters Some of the dynamic properties and parameters of the anti lock brake system can be altered by selecting e Modify Parameters Several brake system related parameters can be changed such as effective rotor radius brake cylinder bore friction coefficient between pad and rotor and the time constant of a first order system used to model the brake system dynamics Driver Model File 41138 Simulation Disturbances Roadway Controllers Scenarios Lane Sensors gt Vehicle gt Tires b finti Lock Brake System Driver 2 Simple Preview Driver Model Path Prediction gt Optimal Preview Driver Model Cruise Control Mouse Driver Driver Status gt No Driver Fixed Steering Wheel Modify Parameters The CAPC simulation tool has two different kinds of driver models built in e Simple Preview Model This model is relatively simple and steers the vehicle by tracking one point at a fixed distance in front of the vehicle The model includes characteristics of the human driver such as neuromuscular related dynamics and time delay The simplicity of the model restricts the application to straight line driving e Optimal Preview Model This model is a more sophisticated version of the simple preview
9. required which includes steering wheel and vehicle speed data as inputs for a maneuver and lateral accelerations and or yaw rate as measured outputs These measured inputs are used as inputs for the vehicle CAPC v4 2 User s Manual 9 simulation within the CAPC tool and the simulated outputs are compared to the vehicle test outputs For example the CAPC project vehicle testing included a set of vehicle handling tests on the test track as reported in LeBlanc et al 1996 The steering and vehicle speed data were then fed into the CAPC tool using the post processor mode and the measured lateral acceleration and yaw rates were compared to the simulated values This allowed us to choose vehicle dynamics parameters such as tire cornering stiffnesses roll damping and stiffness yaw inertia and steering wheel to steering rack model parameters While in post processing mode the user is prompted for the test input file Once this is read in the user begins the simulation and the computer display shows a graphic of a vehicle moving laterally as indicated by the simulation and the vehicle test data We will _ describe this mode in more detail as its full use would require the user to have a full understanding of the vehicle dynamics suspension and steering models Please use the contact information listed at the beginning of this manual The Menus We now describe each of the pull down menus in turn This wiil pr
10. screens according to the seasons of the year All roadway geometries can be combined with any of the seven road unevenness types from the Disturbance menu CAPC v4 2 User s Manual 21 The Controllers Menu Roadway Scenarios Window CAPE on Brake Steer Off Near Field LM Kalman Filter PID Feedback Far Field LM Kalman Filter LOR State Feedback Modify Supervisor Parameters Modify PID Feedback Gains The controller menu controls the specific hardware and software related items for the road departure avoidance CAPC system It has the following items e CAPC The switch enables or disables the entire CAPC system If it 15 switched off the program will simulate only the vehicle dynamics related part including a driver model When on warnings are indicated Intervention is enabled by the Brake Steer option described below e Brake Steer This menu option allows the user to enable or disable brake steer control If it is switched off and the CAPC system switch is on the entire CAPC system will operate in an open loop with only warnings being issued The brakes will not be activated if a road departure 1s sensed when brake steer control has been disabled When Brake Steer is enabled two types of feedback controllers are available a PID feedback of the lateral position error and a more sophisticated linear quadratic state feedback controller which uses the lateral position error heading angle side slip vel
11. software you must have the CAPC v4 2 application program and seven data files which should be packaged with the CAPC and a Macintosh computer There are two versions of the CAPC v4 2 one for older 68K Macintoshes pre PowerPC named CAPC v4 2 68K and one for the PowerPC Macintoshes named CAPC v4 2 PPC The seven data files include an input file named Default STP containing default parameters and six files each containing tables that describe a different road surface unevenness These files are called Faulted PCC Good PCC and SO on For any numerical design or analysis needs you will want to know about the output file format that CAPC v4 2 produces called ERD The CAPC tool saves the user s choice of variables during the course of the simulation This data can be directly plotted using a plotting package named EP or the file headers can be easily stripped off and the data imported into any plotting package or spreadsheet EP is either enclosed with the other files or is available on the Web site at the University of Michigan Transportation Research Institute UMTRD http www umtri umich edu erd_soft ep himl System requirements Apple Macintosh 68K processor with FPU PPC about SMB RAM and 5 MB hard disk space Color monitor 15 or larger recommended but any will do Does not work on most pre 1996 PowerBook laptops 100 200 500 series except those with an FPU PC or UNIX version status Contact Profe
12. OF The sprung mass is able to move in three directions longitudinal lateral and vertical and to rotate about three axes roll pitch and yaw Each wheel suspension has one DOF with respect to the vehicle body and the rotation of a wheel also accounts for one DOF Simpler models are available via pull down menus as described later The tire accomplishes three basic functions 1 supports the vehicle weight cushioning road irregularities 2 develops lateral forces for cornering and 3 develops longitudinal forces for accelerating and braking Several tire models are available in the CAPC simulation tool The most extensive model is based on an empirical model known as the Magic Formula model It describes tire slip forces for a large range of load and slip quantities It also can handle cases of combined slip cornering and braking or accelerating The less extensive models are based on the Magic Formula but contain simplifications to speed computations The CAPC simulation vehicle can be operated on various roadway types with different geometries and unevenness Besides an oval test track a straight road a winding road and a skid pad have been preprogrammed The geometry can be extended with grades and superelevations Furthermore several data sets of different artificially generated road unevenness profiles are available The road friction coefficient can be changed too Two driver models are available in the CAPC simulation tool
13. Test Track Tacom Test Track Dana Test Track Sinus Road 150 Double Lane Change Skidpad Season The roadway menu enables the user to select different roadway geometries The following items can be selected e Straight Road The road is straight and the user is allowed to choose different values for the grade and superelevation e Oval Test Track This roadway geometry corresponds with one lap of a nearby test track oval Both turns have the same radius 381 m and the straightaways are each about 1600 m long This oval contains six lanes each with a different superelevation 0 0 5 2 5 6 15 28 degrees e Short Test Track The geometry of this track corresponds with one lap of another test track in Michigan It has two tight curves 102 m radius 0 degrees superelevation and one larger bend 145 m radius 8 5 degrees superelevation e Steep Test Track The geometry of this track corresponds with the track from yet another track in Michigan It is an oval with two identical bends 279 m radius 6 2 degrees superelevation at the end of the straightaways each 408 m long e Sinus Road This particular roadway has a sine shape with a wavelength of 200 m and an amplitude of 2 m e ISO Double Lane Change This is a free interpretation of the ISO TR 3888 norm for double lane changes e Skid pad The skid pad is a circular track with a radius of 100 m e Seasons The season switch enables the user to change the colors of the animated
14. This screen corresponds with the image as seen through the far field camera The Vehicle windows and the two camera images are updated with different frequencies The camera images are updated with the scanning frequency of the cameras which can be altered in the Models Lane Sensor Modify Parameters menus The two vehicle images are updated with the communication interval as set in the Simulation Timing menu This menu is inactive in the driving simulator or post processor mode References Ervin R D Ed The Crewman s Associate for Path Control CAPC Final report for TACOM Contract DAAE07 93 C R124 University of Michigan Transportation Research Institute Report No UMTRI 95 35 1995 Full description of the implementation of the CAPC road departure prevention system on a prototype vehicle Also describes more about the simulation tool code CAPC v4 2 User s Manual 24 LeBlanc D J Venhovens P J Th Lin C F Pilutti T Ervin Ulsoy G MacAdam C and Johnson G A Warning and Intervention System for Preventing Road Departure Accidents Segel L Ed The Dynamics of Vehicles on Roads and Tracks Proc 14th IAVSD Symposium Ann Arbor MI August 1995 Presents the concept behind the CAPC system primarily algorithms and how they fit together LeBlanc D J Johnson G Venhovens P Gerber G DeSonia R Ervin R Lin C F Ulsoy A G and Pilutti T CAPC A Road departure Preventio
15. a wide range of load slip combinations The output of the model is a longitudinal and lateral tire slip force and the aligning moment The inputs are the longitudinal slip the sideslip angle the vertical tire load and the wheel camber angle The table look up version of the Magic Formula is based on a table that is filled with numbers generated by the original version of the Magic Formula This version is faster because less computations have to be carried out The table look up version doesn t consider combined slip cornering and braking at the same time e Magic Formula Tire Model pure slip This tire model is identical to the look up version as described above It uses the original Magic Formulae and considers only pure slip situations e Magic Formula Tire Model combined slip The combined slip version of the Magic Formula is the most extensive tire model available in this simulation tool Unfortunately the computational burden is significant e Transient Tire Model Selecting this menu item toggles the transient tire model on or off The transient tire model is based on a first order relaxation system with a load dependent relaxation length This option can be combined with any of the five previously described steady state tire models e Modify Parameters Selecting this item will allow the user to modify the tire road friction coefficient CAPC v4 2 User s Manual 14 Anti Lock Brake System Model Simulation Disturbances
16. ances vision system outputs and CAPC control in and outputs The data is stored in an ERD format and can be displayed by a separate public domain engineering plotter It is also possible to look at the simulation in progress real time from the point of view of the driver CAPC v4 2 User s Manual 8 Driving Simulator Mode When using the driving simulator a display similar to that shown below is seen by the user amp File Model Simulation Disturbances Roadway Controllers Scenarios Window warning in di cat or igure 2 Driving Simulator Mode This mode is most often used in conjunction with the computer mouse being defined as the steering input as described below In the figure the selected roadway appears in the windshield and the white hood of the vehicle appears near the bottom The predicted path is seen as a yellow trace ahead of the vehicle When a warning is triggered within the algorithm a yellow arrow appears at the bottom of the screen This arrow is not necessarily a good interface to a real driver but is useful for the purposes of our simulator When an active intervention occurs we use differential braking a red bar near the bottom moves to the left or right indicating the amount of brake pressure applied to the brakes Post processing test data mode This mode is designed primarily for the validation and tuning of vehicle dynamics models using vehicle test data An external test data file is
17. ation Driver Model e Simple preview model with driver limitations time lag UMTRI s optimal preview driver model Cruise Control set resume accelerate Far Field Camera determining the future roadway geometry Near Field Camera determining the heading angle and lateral deviation Path Prediction based on e 2 DOF linear flat vehicle model lateral and yaw DOF Linear tires fixed cornering stiffness Nonlinear tires cornering stiffness as a function of vertical tire load e 3 DOF linear yaw roll vehicle model lateral roll and yaw DOF Linear tired fixed cornering stiffness Nonlinear tires cornering stiffness as a function of vertical tire load Lane Margin Calculation time to lane crossing Driver Status assessment based on vehicle states and steering wheel activation Brake Steer Controller based on LQ feedback of the lateral deviation heading angle side slip velocity and yaw rate CAPC Supervisory Controller based on driver status and time to lane crossing Roll and Pitch angle estimation based on measured suspension deflections Road Unevenness collected from empirical road models Roadway Geometry straight lines curves w lw o superelevations and or slopes Wind Disturbances constant wind crosswind gust random crosswind Real Time Driving Simulator e 8 DOF yaw roll model including wheelspin DOF Magic Formula tires cruise control path prediction lane margin calculation warnings and interv
18. d Front Suspension Modify Aerodynamic Parameters Rear Suspension rer wt The vehicle model menu enables the user to change the type and complexity of the vehicle model and to alter the parameters of this model It contains the following selections e 2 DOF Flat Vehicle Model The 2 DOF vehicle model is characterized by the lateral and yaw DOFs This _ model only be used in the Post processor mode e 7 DOF Flat Vehicle Model The 7 DOF vehicle model is characterized by the following seven degrees of freedom longitudinal lateral yaw and four wheel rotational DOFs e 8 DOF Yaw Roll Vehicle Model The 8 DOF vehicle model is characterized by the following eight degrees of freedom longitudinal lateral roll yaw and four wheel rotational DOFs e 14 DOF Fuli Vehicle Model The 14 DOF vehicle model is characterized by the 6 DOFs for the vehicle body 1 DOF for each axle and four wheel rotational DOFs e Modify Vehicle Parameters This menu item enables the user to change important vehicle parameters such as the masses moments of inertia and vehicle speed of travel e Modify Suspension Characteristics The suspension spring and damping constants can be changed if this is Selected Furthermore it allows the user to change the vertical tire stiffness arid auxiliary roll stiffness of the anti roll bars e Modify Suspension Geometry The geometry of the front and rear suspensions can be altered by selec
19. dn noes Moo We February 1997 The Un igan f Mich iversity Paul J Th Venhovens Transportation Research Institute David J LeBlanc Dept of Mechanical En 108 ng amp Applied Mechani gineeri A igan Research Institute 101 versity of Mich ni Transportat The U Technical Report Documentation Page 1 Report No 2 Government Accession No 3 Recipient s Catalog No UMTRI 97 13 4 Title and Subtitle 5 Report Date February 1997 User Manual for CAPC simulation tool v4 2 6 Performing Organization Code 4 pir a mg p wy PET Tag Ep a D 8 Performing Organization Report No 7 Author s R P Venhovens D LeBlanc Performing Organization Name and Addr 10 Work Unit No TRAIS niversity of Michigan Transportation Research Inst and Dept of Mechanical Engr and Applied Mechanics 11 Contract or Grant No University of Michigan Ann Arbor MI 48109 2150 13 Type of Report and Period Covered 12 Sponsoring Agency Name and Address 14 Sponsoring Agency Code 15 Supplementary Notes 16 Abetract This document is an introductory user s manual for the Crewman s Associate for Path Control simulation tool version 4 2 This user s manual replaces all manuals for previous versions CAPC v4 2 is a Macintosh based software package for simulating a road departure
20. enario except that the road departure will occur in a curve rather than on a straightaway e 2b Curve Pulse Steer This scenario corresponds with scenario 1b except that the road departure will occur in a curve rather than on a straightaway e 3 Curve No Steer CAPC v4 2 User s Manual 23 The steering wheel is fixed in the neutral position and the vehicle is approaching e Modify Scenario Parameters Selecting this item enables the user to change the timing and magnitude of the forced steering wheel input Selecting one of the scenarios will not only initiate a road departure but also other menu options such as roadway and tire model will be altered depending on the scenario chosen With all scenarios the driver model will be disabled once a step pulse steer 15 initiated The Window Menu Scenarios Vehicle 1 Vehicle 11 V Near Field Screen v Far Field Screen With the window menu the user is able to enable or disable the animated screen output of the simulation tool e Vehicle I This window corresponds with the large rectangular window on the bottom of the screen and shows the vehicle roadway and camera ranges e Vehicle This window shows a close up of the vehicle only The red lines on every corner of the vehicle represent the magnitude of the tire slip forces e Near Field Screen This screen corresponds with the image as seen through the near field camera e Far Field Screen
21. ention CAPC v4 2 User s Manual 6 How to Use the Simulation Tool The Three Modes When starting up the CAPC simulation tool the following screen will appear 212279255 e 5 S S SS _ S MASS x SS Se SS SES SS NN 5 S SS SS NS SS SES SNS SEES SESS SESS X S SS 5 Figure 2 CAPC start up screen for an earlier version The simulation tool offers three modes of operation The first mode is the Simulation mode which includes all the features of the lane departure avoidance system The second mode is real time driving simulator mode scaled down version of the ordinary simulation mode Some items have been disabled such as the 14 DOF vehicle model because of the computational burden The driving simulator mode offers an animated view of the vehicle and roadway as seen through the eyes of the driver The third mode is a post processor option that allows the user to input measured steering wheel angle and vehicle speed from the prototype car directly into the simulation The input file needs to be an ERD binary file With specific channel short names The three modes can support different objectives For example the simulation mode is useful for the designer who needs to perform experiments with well known driver inputs to develop or evaluate the road departure system or a subsystem The driving simulator mode which is usually operated using the mouse as a steer
22. he vehicle is based on a simple vehicle model assuming that the steering wheel position and vehicle speed remain constant during the projection The path is obtained by integrating the differential equations of the vehicle Appropriate initial conditions are necessary The following path prediction models methods are available e Steady state Curve The steady state curve prediction 1s based on the steady state solution of the differential equations of a 2 DOF vehicle model lateral and yaw The solution has the shape of a curve with a fixed radius The radius depends on the steering wheel input vehicle parameters and speed of travel e 2 DOF Vehicle Model linear tires The path projection 1s based on a 2 DOF vehicle model lateral and yaw and a linear tire model constant cornering stiffnesses e 2 DOF Vehicle Model nonlinear tires The path projection is based on a 2 DOF vehicle model lateral and yaw and a nonlinear tire model load dependent cornering stiffnesses e 2 DOF Vehicle Model linearized The path projection is based on a 2 DOF vehicle model lateral and yaw and a linear tire model constant cornering stiffnesses The geometry related to the path prediction has been linearized cos V 1 sin w y e 3 DOF Vehicle Model linear tires CAPC v4 2 User s Manual 16 The path projection is based on a 3 DOF vehicle model lateral roll and yaw and a linear tire model constant cornering stiffnesses e 3 DOF Vehicle Model n
23. hicle Even the speed of the vehicle itself does not generate aerodynamic forces e Continuous Wind There will be a continuous wind speed present The direction and the magnitude of the wind speed can be altered in the Modify Wind Parameters menu item The orientation of the wind 1s assigned in an inertial frame The default wind speed is zero e Crosswind Gust The gust is a side wind speed with an orientation according to the y axis of the inertial coordinate system The gust is of a pulse shape The start time stop time and DC wind speed can be altered by selecting the Modify Wind Parameters menu item e Random Crosswind The random crosswind gust is identical to the Crosswind Gust as explained above except that the wind speed is randomly distributed The random speed effect has been obtained by low pass filtering of white noise In addition to the start and stop time the standard deviation and frequency contents of the random component can be changed by selecting the Modify Wind Parameters menu item e DC Random Crosswind This option is identical to the Random Crosswind except that the random wind speed 1s superimposed on a DC side wind speed e Modify Wind Parameters Depending on the type of wind disturbances chosen selecting this item will allow the user to modify particular wind parameters such as speed and pulse duration times CAPC v4 2 User s Manual 20 The Roadway Menu EUMENE Controllers Scenarios Straight Road Qual
24. implies that the images of the cameras are affected by the vehicle motions These motions include body roll during cornering pitch during braking or acceleration or a combination of heave roll and pitch resulting from operation on an uneven road The transformation from screen coordinates to roadway x y coordinates can be adjusted 1f the motion of the vehicle is known Two options are available e No Compensation No compensation means that the motion of the vehicle 1 not included in the transformations from image coordinates to roadway geometry coordinates e Roll amp Pitch Compensation correction will be applied in the transformation based on estimates of the roll and pitch angle of the vehicle determined by suspension deflection measurements The last submenu item 15 e Modify Parameters Selecting this menu item enables the user to modify the position and orientation of the cameras on the vehicle Furthermore the focal length image update rate and the range of interest can be changed CAPC v4 2 User s Manual 12 Vehicle Model File KULE Simulation Disturbances Roadway Controllers Scenarios Window 2 DOF Flat Vehicle Mode DOF Flat Vehicle Model 8 DOF Yaw Roll Vehicle Model 14 DOF Full Vehicle Model Tires Anti Lock Brake System Driver Path Prediction Cruise Control Modify Vehicle Parameters Driver Status Modify Suspension Characteristics _ Modify Suspension Geometry i
25. ing wheel is useful to both demonstrate the system or subsystem but also to test behavior when driver steering inputs are closer to real For instance an error in our first warning logic was uncovered with this mode the error appeared only when the center line had been crossed twice Driving for five minutes allows the engineer to quickly test many scenarios CAPC v4 2 User s Manual 7 The third mode is useful for comparing data measured in real vehicle with the simulation computations This was used to tune vehicle dynamics and tire dynamics models and to calibrate the model for our steering system dynamics Simulation Mode The simulation tool is menu driven and is supported by various animations The user is able to modify all important model parameters using pull down menus and dialog boxes The simulations are supported by graphical and numerical outputs The graphical animation shows the vehicle from a top view including the roadway geometry predicted future trajectory and perceived roadway geometry The scenery of the roadway as recorded by both vision systems is also displayed during the animation Figure 1 gives an impression of the animation File Model Simulation Disturbances Roadway Controllers Scenarios Window CAPC Animation Near field Camera Far field Camera CAPC Animation Figure 1 CAPC simulation The numerical output contains all important states of the vehicle roadway geometry disturb
26. m will quit the simulation program The Models menu The models menu allows the user to select and modify items related to the dynamics models Different models for the vehicle tires driver and path prediction can be selected The complexity of the model varies with the selection and it is also possible to change specific parameters of each model The models menu contains eight items vision system vehicle models tire models antilock brake system models driver models path prediction models cruise control model and driver status model Each of these is now described CAPC v4 2 User s Manual 11 Vision System File Simulation Disturbances Roadway Controllers Scenarios Window Lane Sensors M Near field Camera 4 Uehicle No Compensation Tires amp Pitch Compensat Anti Lock Brake System asd in p ion b b Driver Modify Parameters gt 4 Path Prediction Cruise Control Driver Status The CAPC vehicle is equipped with two vision systems digital video cameras e Near Field Camera The near field vision system is scanning the roadway geometry close in front of the vehicle e Far Field Camera The far field vision system is also scanning the roadway geometry in front of the vehicle However the range is longer than that of the near field camera Both vision system related menus come with a submenu The cameras are fixed on the vehicle body This
27. model It minimizes the tracking error at several points in front of the vehicle This model is better for complicated maneuvers and driving through curves The model also includes a pure time delay Besides the two driver models the following options are available e Mouse Driving The vehicle can be steered by moving the computer mouse from left to right The neutral position corresponds to the middle of the screen e No Driver Fixed Steering Wheel CAPC v4 2 User s Manual 15 As the name already suggest the steering wheel is fixed in the neutral position with this option The characteristic parameters of the driver model can be altered by selecting e Modify Parameters Depending on the driver model chosen simple optimal preview mouse driver the selection will enable the user to modify time constants and gains Path Prediction Model File 138 Simulation Disturbances Roadway Controllers Scenarios Window Lane Sensors Vehicle Tires Anti Lock Brake System Driver Path Prediction Steady State Curve Cruise Control Driver Status 2 DOF Flat Vehicle Model linear tires 2 DOF Flat Vehicle Model non linear tires 2 DOF Flat Vehicle Model linearized 3 DOF Yaw Roll Model linear tires 3 DOF Yaw Roll ug linear tires Modify Parameters The time to lane crossing algorithm as used in the CAPC program uses the future vehicle trajectory as an input to the calculations The future path of t
28. n System IEEE Control Systems Magazine Vol 16 No 6 pp 61 71 December 1996 Presents the prototype implementation of the CAPC system which differs some from the algorithms of the LeBlanc et al 1995 paper LeBlanc D J Johnson G Venhovens P Gerber G DeSonia R Ervin R Lin C F Ulsoy A G and Pilutti An Implementation of Road departure Warning System Proc 1996 IEEE Intl Conf on Control Applications pp 590 595 Dearborn MI September 1996a Earlier version of the LeBlanc et al 1996 paper Lin C F Lane Sensing and Path Prediction for Preventing Vehicle Road departure Accidents Ph D dissertation The University of Michigan 1995 Ph D dissertation with detailed description and justification for a set of time to lane crossing algorithms including a far range Kalman filter and another option not implemented in the simulation tool Papers with first author Lin below address various topics related to this Lin C F and Ulsoy G Calculation Of The Time To Lane Crossing And Analysis Of Its Frequency Distribution Proc of American Control Conference Seattle WA 1995 Lin C F and Ulsoy G Vehicle Dynamics And External Disturbance Estimation for Vehicle Path Projection Proc of the American Control Conference Seattle WA 1995a Lin C F Ulsoy G LeBlanc D J Lane Geometry Reconstruction Least Square Curve Fit Versus Kalman Filter Proc ASME Intl Congres
29. n the simulation and in LeBlanc et al 1996 The Scenarios Menu Controllers Window No Scenario 1a Straight Road Step Steer 1b Straight Road Pulse Steer 2a Curve Step Steer 2b Curve Pulse Steer 3 Curve No Steer Modify Scenario Parameters The scenario menu contains some pre programmed scenarios that will cause a road departure due to a forced steering input The following selections can be made e No Scenario No additional steer input will be generated if this menu item is selected The driver input whether through a model see Model menu or the user using a mouse is used directly la Straight Road Step Steer A step steer input will initiate a road departure The magnitude and timing of the step can be altered in the Modify Scenario Parameters menu item The simulation begins with the default driver model steering the simulated vehicle At the appropriate timing as defined by default or by the user in the Modify Scenario Parameters menu item a step input of steering wheel angle is input The steering wheel will remain in the non neutral position 1b Straight Road Pulse Steer A pulse steer input will initiate a road departure The magnitude and width of the pulse can be altered in the Modify Scenario Parameters menu item The steering wheel will be in the neutral position again after the pulse has been applied e 2a Curve Step Steer This scenario corresponds with sc
30. ocity yaw rate and integral of the lateral position error as inputs The gains of the PID feedback controller can be changed in the Modify PID Feedback Gains option The LQR gains are hard coded in the simulation software as a function of the vehicle speed The PID design 15 described in Pilutti and Ulsoy 1995 and the LQR design described in LeBlanc et al 1995 e Near Field LM Kalman Filter Switching this option on means that a Kalman filter will be used to match measured near field lane marker LM data with a roadway geometry model and thus to filter out noise If the Kalman filter is disabled least square curve fitting will be used instead The Kalman filter is described in LeBlanc et al 1995 e Far Field LM Kalman Filter Switching this option on means that a Kalman filter will be used to match measured far field lane marker LM data with a roadway geometry model If the Kalman filter is disabled least square curve fitting will be used instead These methods are described in Lin et al 19955 e Modify Supervisor Parameters The supervisory controller decides whether or not to warn intervene if a road departure is detected By selecting this option the user can change the sampling CAPC v4 2 User s Manual 22 rate and some threshold values related to the decision making module The decision to warn and or intervene uses thresholds on time to lane crossing and vehicle lateral position as described in both a dialog box i
31. oftware example of a system design problem was study to evaluate whether the roadway geometry estimation algorithm would work on a steeply banked test track that was available for our early vehicle testing This required a simulation tool with the combination of roadway and vehicle modeling vision sensor and estimation modeling as well as the ability to compute the metrics and logic used to initiate warnings to the driver The finding was that while Kalman filters ameliorate the problem of large superelevation there is a tradeoff between obtaining good performance in the turn versus good performance approaching the turn essentially the weighting of vision data against vehicle motion data The simulation tool has played a significant role as the CAPC system grew to its final hardware design Before testing on the road or proving grounds control strategies have CAPC v4 2 User s Manual Eds been implemented in the simulation tool to verify their efficiency and stability The ability to exchange code has simplified and shortened the implementation phase significantly Dynamics Models The core of the simulation is the vehicle models While driving on the road the vehicle is subjected to many external inputs such as driver steering wheel actions road unevenness and wind The CAPC simulation tools offers a variety of vehicle models each designed for a specific application The most extensive model contains 14 degrees of freedom D
32. onlinear tires The path projection is based on a 3 DOF vehicle model lateral roll and yaw and a nonlinear tire model load dependent cornering stiffnesses Cruise Control File Mur Simulation Disturbances Roadway Controllers Lane Sensors Vehicle Tires Anti Lock Brake System Driver Path Prediction Cruise Control 21 Cruise Off Driver Status bi Cruise On Accelerate Modify Parameters rr The cruise control has three modes of operation e Cruise Off The cruise control is switched off e Cruise On The cruise control is switched on and holds the forward speed of the vehicle constant e Accelerate The cruise control is switched on and the controller follows a predefined speed pattern The speed pattern can be modified by selecting the Modify Parameters menu item Driver Status Model File 1 Simulation Disturbances Roadway Controllers Lane Sensors Vehicle Tires Anti Lock Brake System Driver Path Prediction Cruise Cantrol Driver Status Modify Parameters The status of the driver can be an input for the CAPC decision module The algorithm monitors several vehicle roadway related signals such as lateral deviation of the vehicle and steering wheel input The moving average and standard deviation values are combined to one quantity denoted by PERCLOS the proportion of the time that the driver s eyes are 80 to 100 percent clo
33. ool the default vehicle parameters will be loaded from the file Default STP Do not erase this file or hundreds default parameter values will be lost This default file contains information which approximates a 1995 Ford Taurus SHO the vehicle used to implement the CAPC road departure prevention system LeBlanc et al 1996 e Load Simulation Setup A previously saved setup can be loaded by selecting this item the current vehicle and simulation related parameters will be replaced by the data from the setup file e Load ERD test file This option is only selectable in the Post processor mode of the CAPC simulation tool A file in an ERD format with specific short names will be used to run a simulation in a post processor mode This file must be recorded with the CAPC prototype vehicle and MacDAS software The front wheel steer angle _ and vehicle speed as stored in the ERD file are ported directly into the simulation tool such that a simulation can be carried out with measured driver inputs The ERD file must be in a binary floating point format and needs to contain the following short names d_sw steering wheel angle in degrees d_fw front wheel steer angle in degrees u_RF right front wheel speed in m s u_LF left front wheel speed in m s u_RR right rear wheel speed in m s u_LR left rear wheel speed in m s The vehicle speed is derived by averaging all four wheelspin derived speeds e Quit Selecting this menu ite
34. ovide a good feel for what features are available and give the user a basic how to description The menus are generally very similar in the different modes Differences in the menus when using the other modes are straightforward and do not require separate discussion The File menu WT Model Simulation Disturbances Modify ERD Qutput Setup Save Simulation Setup Load Simulation 0 AAA NAA ANNA AR A ABA AAA NA FRAIS Load ERD test file The file menu enables to user to modify the simulation It contains the following menu items e Modify ERD Output Setup The user may select and modify the variables to be saved for plotting after simulation run The output selection can be done by selecting this item A dialog will appear with a large amount of check boxes Choose the signals that you would like to study after a simulation run The data will be stored in a text ERD format and the signals can be viewed using the ERD plotter e Save Simulation Setup CAPC v4 2 User s Manual 10 the parameters of a specific vehicle are stored in a setup file STP Besides vehicle parameters the setup file contains simulation related parameters such as time step road unevenness and controller gains After modifying the vehicle and simulation related parameters the setup can be saved by selecting this menu item The setup file needs to have the extension STP When starting up the simulation t
35. s and Exhibition San Francisco CA 1995b Lin C F Ulsoy A G LeBlanc D J Lane Geometry Perception and the Characterization of its Associated Uncertainty First submitted to the ASME Journal of Dynamic Systems Measurement and Control Aug 5 1995 resubmitted Sep 14 1996 Lin C F and Ulsoy G Time to Lane Cross Calculation and Characterization of its Uncertainty 175 Journal vol 3 no 2 pp 85 98 1996a Pilutti T and Ulsoy A G Vehicle Steering Intervention Through Differential Braking Proc of American Control Conference Seattle WA 1995 First work to explore design and capability of closed loop control on brake pressure to control lateral motion See LeBlanc et al 1995 for a next generation controller Venhovens J Th Reference Manual For Road Departure Prevention System Simulation Program The University of Michigan Transportation Research Institute 1995 Manual describing dynamic models of the CAPC simulation tool About 100 pages CAPC v4 2 User s Manual 25
36. sed a quantity described in technical reports and papers from W W CAPC v4 2 User s Manual 17 Wierwille at Virginia Tech The regression parameters can be changed by selecting the Modify Parameters menu item Currently the driver status is not used in the decisions to warn and or intervene to prevent road departure The Simulation Menu Model Simulation 8990422409840 44604409 480984410040 0008628 80 2 0849 90 44 00410 0806000449 esed saso aed tese qun cat 054004 eres o Gua Euler 1st order Runge Kutta 2nd o Runge Kutta 4th o Neb 980000 00 0000999080004 0004400000088 5800 8200400 0 20 SP 12809 04 209 R30 FA 4C UPS 40040 Continue Reset Pulling down the simulation menu enables the user to change the tool mode start a simulation or change integration related settings such as the time step or duration The following menu items are selectable e Choose Mode Selecting this item returns the user to the high level three mode choice shown earlier in Figure 2 The mode choices are Simulation Driving simulator or Post processing mode e Timing Selecting the timing option enables the user to change the simulation time step duration and communication interval The communication interval determines the output to the ERD data file and to the screen e Euler Ist order
37. ssor Huei Peng at the University of Michigan if you must have an IBM compatible or a UNIX compatible version He s developed a version for the graphics library OpenGL This is available albeit with fewer capabilities than that of the Macintosh version and is aimed at intelligent cruise control uses CAPC v4 2 User s Manual 2 More information on the software A Reference Manual was written by P J Th Venhovens in 1995 This 100 page document describes in detail the dynamics models See the contacts below to obtain this For more information Contacts Director of the CAPC projects Mr Robert E Ervin Head Engineering Research Division The University of Michigan Transportation Research Institute UMTRI 2901 Baxter Rd Ann Arbor MI 48109 2125 USA Phone 313 936 1066 Email bervin Q umich edu Resident software developer Mr Kevin O Malley UMTRI Email omalley umich edu Manuals and papers See the Web site http www personal engin umich edu leblanc dir research ActiveSafety html or contact Professor A Galip Ulsoy at the University of Michigan CAPC v4 2 User s Manual 3 Description of the CAPC v4 2 Tool Introduction The different subsystems of a lane departure avoidance system have been combined on a simulation level in the CAPC Crewman s Associate for Path Control simulation tool The simulation tool is a modular concept written in C language and runs on Macintosh computers using either a PowerPC
38. ting this item The geometry of the true suspension has been represented by a simplified suspension model Modifying the parameters will affect the track width and roll center height e Modify Aerodynamic Parameters The six aerodynamic coefficients of the vehicle body can be changed when this item 15 selected Furthermore the frontal area of the car the barometric pressure and air temperature can be altered CAPC v4 2 User s Manual 13 Tire Model File Simulation Disturbances Roadway Controllers Scenarios Window Lane Sensors Vehicle Tires 2 Cornering Stiffness constant Anti Lock Brake System Cornering Stiffness load dependent Driver gt Table look up based on Magic Formula Path Prediction gt Magic Formula Tire Model pure slip Cruise Control gt Magic Formula Tire Model combined slip Driver Status v Transient Tire Model Modify Parameters The tires are separated from the vehicle menu item The following slip force models can be chosen e Cornering Stiffness constant This tire model is the simplest model available It is using constant slip stiffnesses to calculate the tire slip forces and moments e Cornering Stiffness load dependent This tire model takes the load dependency of the slip stiffnesses into account while calculating the tire slip forces e Table Look up Based on Magic Formula The Magic Formula is an empirical tire model that 1s valid for
Download Pdf Manuals
Related Search