A Hardware-in-the-Loop and Virtual Reality Test Environment for Steer-by-Wire System Evaluations Pradeep Setlur, Dr John Wagner‡, Dr Darren Dawson, and Lance Powers Automotive Research Laboratory Departments of Mechanical and Electrical/Computer Engineering Clemson University, Clemson, South Carolina 29634 Abstract A high precision, cost effective, experimental hardware-inthe-loop steer-by-wire test environment is presented and discussed to support engineering and psychology studies In this project, a suite of chassis models and nonlinear control algorithms are developed and validated, as well as various human-machine interface design issues investigated This paper provides an overview of the real time steering simulator which has been created to facilitate the comparison and evaluation of vehicle and steering system control strategies in a repeatable manner The insertion of novel driver input devices, with adjustable force feedback, permits the study of the human-vehicle interface The remote operation and/or supervision of semi-autonomous or autonomous vehicles can be studied using a human-in-theloop system to provide insight into the relative importance to the driver of various vehicle parameters Introduction The emergence of unmanned/manned ground vehicles represents an important tool in the manufacturing environment, intelligent highway system, space exploration, defense, and security fields to complete repetitious and hazardous tasks An important vehicle technology is driveby-wire [1] which permits local and remote operation in both semi-autonomous and autonomous scenarios The efficient development of steer-by-wire hardware and software for both the human-machine haptic interface and the directional control assembly servo-motors require a repeatable test environment, similar to the actual vehicle, with a variety of prescribed operating profiles One of the advantages of laboratory testing is the ability to fully address safety and performance issues prior to extensive invehicle testing In this paper, an unmanned/manned vehicle steering system hardware-in-loop testbed with virtual reality capabilities will be presented The vehicular steering testbed components include the steering and chassis mathematical models, driver interface with adjustable force feedback, rack and pinion directional control assembly, tire/road interface, and driver With the help of various vehicle databases, the hardware-in-the-loop experimental station can facilitate the investigation of different hybrid vehicle ‡ steering strategies Although the initial focus of the project is on front wheel steer-by-wire technology for ground transportation vehicles, the emerging concept of four wheel steer-by-wire systems can be accommodated with the appropriate mathematical models, control algorithms, and steering hardware The scenario of autonomous vehicle operation requires a tracking controller that can follow a prescribed path or trajectory One possible operating mode is for the human operator to supervise the vehicle's functionality and intervene, if appropriate, to assume control of the steering actions An important consideration is the availability of reliable information that can pinpoint the vehicle's orientation and position The haptic interface utilizes a nonlinear tracking controller to ensure that the steering mechanism follows the operator commands and simultaneously, a tunable force feedback is provided to the driver The driver's interface features a switched reluctance motor to offer resistive feedback for the steering maneuver based on the desired road "feel" A high torque dc motor in the directional control assembly actuates the front wheels to the desired steer angle To provide the driver with visual feedback during the steering operation, a virtual reality tool has been integrated to display the changing environment as a result of the vehicle's motion Standard steering maneuvers may be completed to evaluate the performance of the control algorithms The paper is organized as follows In Section 2, the modeling and control algorithms for the steering subsystem are discussed The hardware-in-the-loop steer-by-wire test platform is described in Section Preliminary experimental results are presented in Section to demonstrate the opportunities to impact the drive-by-wire design process Finally, concluding remarks are discussed in Section Modeling and Control of Steering Systems The foundation of the steer-by-wire test environment begins with the mathematical modeling of the steering system, chassis, tire/road interface, and wheels (refer to Figure 1) Servo-motors exist within the driver interface and (front wheel) directional control assembly to replace the functionality of the traditional hydraulic pump with spool valve and mechanical linkage between the steering column Corresponding author: jwagner@clemson.edu 0-7803-7896-2/03/$17.00 ©2003 IEEE 2584 Proceedings of the American Control Conference Denver, Colorado June 4-6, 2003 and rack and pinion The mathematical models are needed to recreate a realistic steering feel to drivers in the steer-bywire system To regulate the motor voltages, real time control algorithms must be designed and implemented with Steering Control Unit Driver Interface and Directional Control Assembly experimental validation to be performed on the steering test platform The four wheel steer-by-wire system requires at least one additional servo-motor to regulate the rear rack and pinion directional control assembly Steering Linkages Steer-by-Wire System Environment Chassis System Powertrain System Engine Control Unit Driver Engine/ Transmission Brakes Wheel Dynamics Tire/Road Interface Platform Dynamics Suspension System Figure 1: Vehicle dynamics diagram for a steer-by-wire system 2.1 Steer-by-Wire Vehicle Dynamics The emergence of hybrid vehicles, which may include a gasoline or diesel engine, electric motor with battery pack, and/or fuel cell, to achieve fuel efficiency and emissions gains, mandates alternative power steering systems An attractive replacement to the traditional hydraulic steering system assist is steer-by-wire, which replaces the hydraulic fluid actuation with a high torque servo-motor The direct link between the driver and wheels has been removed and replaced by a shortened steering column which is connected to a low torque servo-motor The automotive steering system has been investigated for the last five decades to provide improved vehicle lateral performance and insight into the underlying dynamics for enhanced steering system designs Analytical and empirical dynamic models allow the driver's commanded steering wheel input to king pin torque output to be studied in terms of the vehicle's lateral motion Post and Law [2] developed a hydraulic power steering system model to evaluate a vehicle’s lateral response to various driving maneuvers Electrical [3] and steer-by-wire [4] power assist steering system configurations have been examined to investigate the handling gains and safety requirements To facilitate the investigation of the human-machine interface and vehicle's lateral responsiveness to steer-bywire systems, a suite of nonlinear models have been developed In the paper by Mills et al [5], the governing dynamics are derived and presented for hydraulic, electric, and steer-by-wire steering systems Of particular interest are the steer-by-wire subsystem equations based on the freebody diagram shown in Figure This steering model has been integrated with a fourth-order chassis model (e.g., [6]) to estimate the vehicle behavior for various driving maneuvers in the Matlab/Simulink software environment Initially, a passenger vehicle was modeled and the simulation results undergoing validation with experimental transient response data The equation of motion developed for the front wheel steer angle [5] is of interest in studying the effects of the vehicle dynamics on the driver-vehicle interface These lumped parameter dynamics are described by y δ&&i = − K L δ i − rack rL Iw − T fr ,kp − Bkpδ&i − M zi − I wω i Ω i i where T fr ,kp is the nonlinear king pin friction [7], M z is the aligning torque at the road/wheel interface, Bkp in the front wheel assembly damping coefficient, and I wωΩ is the gyroscopic torque about the steering axis with precession angular speed Ω = f ( δ ,δ& ,γ ,γ& , z ) which is a function of wheel camber angle, γ , and wheel bounce, z, respectively A four wheel steering system introduces a rear steering actuator (e.g., second rack and pinion) or individual distributed actuators at each wheel for enhanced vehicle maneuverability and responsiveness in comparison with the 2585 Registered trademark of The Mathworks, Natick, MA Proceedings of the American Control Conference Denver, Colorado June 4-6, 2003 conventional front wheel steering system [8] For example, the rear assembly may be counter-phased at low speeds for greater maneuverability and same-phased at high speeds for improved stability Lee et al [9] investigated the vehicle handling performance available with independent four wheel steering systems trajectory to follow a given reference path This functionality is required for autonomous vehicle operating modes The subsequent Lyapunov-based stability analysis demonstrated that the position and orientation tracking errors were globally, exponentially forced to a neighborhood about zero, which can be made arbitrarily small 2.2 Autonomous Vehicle Controller Extensive numerical simulations have been performed in both SIMNON (SSPA, Department of Automatic Control, Lund Institute of Technology, Sweden) and Matlab/Simulink to validate the controller design At this time, various mathematical models from different sources have been selected to generate the vehicle dynamics This modeling constraint may be partially attributed to the real time execution requirements within the experimental test station Thus, the validation efforts take on greater emphasis to ensure that the system performance is acceptable especially in the presence of unmodeled dynamics In Section 3, the hardware-in-the-loop test platform discussion provides an overview of the important (i.e., primary) components of the infrastructure needed in such a testing facility As explained in Section 2.1, an electric or steer-by-wire system can satisfy the variable power source cycling scenarios Although electric steering is an acceptable alternative, it does not permit autonomous vehicle operation due to the required driver's steering wheel input In contrast, steer-by-wire systems can accommodate both semiautonomous and autonomous operating modes which facilitates the deployment of these technologies Reliable positioning systems (e.g., global positioning systems) enable accurate trajectory generation and following since the error between the actual and desired paths may be continually monitored for corrective actions In the paper by Setlur et al [10], an exact model knowledge nonlinear tracking controller has been designed to force the vehicle's Driver interface Control motor I sw Vs R L K S1 I M1 θM1 θM2 Bsc F fr , rack BM L R ia1 IW Vs1 T fr ,kp BM Brack mrack y rack KL Bkp K s2 IM2 ia Steering linkages Bkp T fr ,kp KL IW Mz δF δF Mz Figure 2: Steer-by-wire system mechatronics diagram 2.3 Haptic Interface Controller The natural progression in this research leads to the requirement of road "feel" provided to the driver in both the cases of local and remotely operated vehicles The need for force feedback has been extensively studied and understood in the field of tele-operated robotics As reported by Liu and Chang [11], force feedback is the second highest rated inputs to the driver (note that vision ranks highest) While force feedback provides the driver with an invaluable sensory input, excessive feedback will expose the operator to unnecessary vibrations and road surface disturbances (e.g., bumps) Thus, it is essential for the control strategy to ensure that the road "feel" provided by the force feedback can be adjusted to provide a comfortable driving environment Many researchers have worked to establish a model for generic systems and performed experiments to identify system parameters with the intention of providing simulated force feedback (e.g., [12] and [13]) Recently, Setlur et al [14] developed a method for integrating force feedback into the driver interface without compromising closed-loop stability The impedance control formulation [15] uses a target system to generate the 2586 Proceedings of the American Control Conference Denver, Colorado June 4-6, 2003 reference signal for the driver interface The controller adapts for parametric uncertainties in the system while ensuring global asymptotic tracking for the "driver experience error'' and the "locked error'' However, torque measurements are required An extension was presented to eliminate the torque sensor measurements, albeit knowing the system parameters To illustrate the strategy, assume that the simplified steer-by-wire driver interface system dynamics are ( ) ( ) θ&&1 + N θ ,θ&1 = τ + T1 , θ&&2 + N θ ,θ&2 = τ + T2 where θ (t ) and θ (t ) represent the angular position of the driver input device and vehicle steering system, respectively The auxiliary nonlinear functions N1 (.) and N (.) describe the driver side and the mechanical system dynamics The variables τ ( t ) and τ ( t ) denote the driver input torque and transmission torque between the actuator on the steering column and mechanical subsystem actuated by the steering column Finally, T1 ( t ) and T2 (t ) are the control input torque applied to the driver input device and the steering column, and the required torque at the wheel/road interface The constant values have been assumed to be unity without loss of generality Then θ&&1 is forced to follow reference θ d which is generated based on θ&& + N ( θ ,θ& ) = α τ + α τ to ensure appropriate d T1 d d 1 2 force feedback Lastly, θ is forced to follow θ such that the vehicle is steered as commanded by the driver It is important to note that pure numerical simulations fall short in providing complete insight into the efficacy of the controller performance since "feel" must be experimentally demonstrated which necessitates experimental testing To facilitate the investigation of alternative driver feedback variables, based on the available vehicle dynamics, in the steer-by-wire system, an index will be constructed that incorporates scaled torque information For instance, the index may be composed of the aligning torques, king pin friction torques, and gyroscopic torques, as well as additional data such as roadway obstacles, surrounding vehicle locations, and environmental hazard information The individual elements may be toggled on/off and scaled to achieve the most desirable feedback to the specific driver through experimental testing A real time hardware-in-theloop (e.g., [17]) experimental test platform has been constructed to provide a virtual reality interface for operators to drive the "vehicle" and evaluate various steering input devices (refer to Figure 3) The experimental work has been performed in the Automotive Research and Mechatronics Laboratories at Clemson University Pentium III-based computer workstations, executing realtime LINUX operating system and Matlab/Simulink based models, provide the means to implement the control algorithms Additionally, a graphical user interface, developed in-house, on the QNX operating system using C++ has also been used to demonstrate the flexibility of the testbed Data acquisition and control implementation are performed using a Servo-To-Go input/output board at a frequency of up to 800HZ The various displacements in the steering system are measured directly from encoders and/or LVDT transducers (refer to Figure 4) Furthermore, torque transducers are used to measure the user's applied and the tire/road interface torques A switch reluctance motor (SRM) along with its torque driver provides the necessary force feedback to the driver A servo motor attached to the steering column is used to provide force feedback and/or steer the vehicle Additional active and passive devices have been added at the rack to apply different road reaction forces A second computer workstation performs virtual reality (VR) computations which can be performed in Matlab However, to facilitate the need for creating custom scenes and situations, VRML rendering was considered more fruitful A 60" X 80" screen, along with a high capacity projector, provide the visual feedback for the driver-in-the-loop experiments Hardware-in-the-Loop Testbed Figure 3: Steer-by-wire test bench with computer control A variety of driving simulators are currently used in the automotive industry including the motion based VIRTTEX simulator at Ford [16] and assorted fixed based commercial units (e.g., DriveSafety by GlobalSim, STISIM Drive by Systems Technology) In this project, a cost effective test station has been developed which allows investigators to focus on the human-machine interface of steer-by-wire systems A suite of mathematical models, nonlinear control algorithms, and steering interface designs may be evaluated An important feature of the research is alternative steering input devices [18] that may assist both normal and physically challenged drivers For example, custom designed joysticks which integrate force feedback have been designed to accommodate individual drivers (refer to Figure 5) It is anticipated that the research will yield appropriate feedback metrics and mechanism designs for all 2587 Proceedings of the American Control Conference Denver, Colorado June 4-6, 2003 types of drivers working in both local and remote vehicle operating modes The steering test platform will also support psychology studies on the operator's decision making process and ability to handle driving distractions the steering wheel in the testbed has been displayed In this experiment, the operator has been requested to perform a standard "J" turn maneuver; haptic force feedback was not initially considered However, various force feedback situations shall be introduced for the haptic interface design studies as previously discussed Hydraulic Cylinder Mobile Bench 20 LVDT Rack Motor # Torque Sensor # Motor # Amp #1 (+/- 15 VDC) Power Supply -40 Optical Encoder # Amp #2 Optical Encoder # (+ 90 VDC) -20 (+ 24 VDC) Power Manifold y direction (meters) Torque Sensor # -60 -80 -100 Optical Encoder # Output -120 -140 Torque Sensor # Output Torque Sensor # Output -160 Optical Encoder # Output I/O BOX Servo Amplifier -180 LVDT Output 20 40 60 80 100 x direction (meters) 120 140 160 Figure 6: Simulated vehicle trajectory in response to user torque at the steering wheel PC Figure 4: Experimental test station component schematic To accommodate four wheel steer-by-wire systems, an additional number of steering actuators are required For coupled front/rear steering, two rack and pinion assemblies with high torque dc motors are needed Similarly, independent four wheel steering requires four dc motors that are distributed to each corner of the vehicle In each case, the driver's interface remains the same However, the steering control algorithm, which regulates the motors, must be redesigned to realize the desired wheel steer angles Several typical driving scenarios will be created and simulated using the test platform Some of the research studies will include i) exploring the steering system's “feel” as communicated to the driver, ii) developing the haptic interface force feedback index, and iii) semi-autonomous and autonomous vehicle operation from a coupled controls and operator perspective in normal/degraded conditions Conclusions Figure 5: Two prototype force feedback driver interfaces In this paper, a hardware-in-the-loop vehicle simulator has been presented to support steer-by-wire system development and human-machine interface studies The overall concept of the test station is shown in Figure which features a Honda CRV interior to house the steer-bywire simulator's hardware/software and create a realistic, yet safe, environment Preliminary test results along with the associated theory have been briefly discussed While the computational power of the PC was sufficient for individual tests, to provide testing facilities for driver-in-the-loop experiments, a dSPACE based system will be used to control the hardware and virtual reality functions Experimental Results Acknowledgement Preliminary tests are being performed to validate the experimental steer-by-wire laboratory test platform The simulated vehicle's response has been compared with a conventional hydraulic assist steering system passenger vehicle since test data exists for this configuration However, additional validation activities are planned using light and medium-duty vehicle test data The vehicle tracking controller has been validated numerically; the haptic interface is undergoing operator-in-the-loop testing As shown in Figure 6, the Cartesian trajectory of the simulated vehicle in response to the user's provided input at The authors gratefully acknowledge the partial support of this work by the U.S Army Tank Automotive Command through the Automotive Research Center References [1] 2588 Kelling, N., and Leteinturier, P., "X-by-wire: Opportunities, Challenges and Trends", SAE paper no 2003-01-0113, 2003 Proceedings of the American Control Conference Denver, Colorado June 4-6, 2003 [2] [3] [4] [5] [6] [7] [8] [9] Post, J W., and Law, E H., "Modeling, Simulation and Testing of Automobile Power Steering Systems for the Evaluation of On-Center Handling", SAE Paper No 960178, 1996 [10] Setlur, P., Dawson, D., Wagner, J., and Fang, Y., Milsap, S., and Law, E., “Handling Enhancement due to an Automotive Variable Ratio Power Steering System Using Model Reference Robust Tracking Control”, SAE paper no 960931, 1996 [11] Liu, A., and Chang, S., "Force Feedback in a Harter, W., Pfeiffer, W., Dominke, P., Ruck, G., and Blessing, P., “Future Electrical Steering Systems Realizations with Safety Requirements”, SAE paper no 2000-01-0822, 2000 [12] Ryu, J., and Kim, H “Virtual Environment for Mills, V., Wagner, J., and Dawson, D., “Nonlinear Modeling and Analysis of Automotive Steering Systems for Hybrid Vehicles”, proceedings of the ASME IMECE, Design Engineering, NY, NY, 2001 Xia, X., and Law, E H., “Nonlinear Analysis of Closed Loop Driver/Automobile Performance With Four Wheel Steering Control”, SAE paper no 920055, 1992 Post, J W., "Modeling, Simulation, and Testing of Automobile Power Steering Systems for the Evaluation of On-Center Handling", Ph.D dissertation, Department of Mechanical Engineering, Clemson University, 1995 Chong, U., Namgoong, E., and Sul, S., “Torque Steering Control of 4-Wheel Drive Electric Vehicle”, proc IEEE Workshop on Power Electronics in Transportation, pp 159-164, Dearborn, MI, 1996 Lee, S., Lee, U., Ha, S., and Han, C., “Four-Wheel Independent Steering System for Vehicle Handling Improvement by Active Rear Toe Control”, JSME International Journal, Series C, vol 42, no 4, pp 947-956, 1999 "Nonlinear Tracking Controller Design for Steer-byWire Automotive Systems", proc American Control Conference, Anchorage, AK, pp 280-285, 2002 Stationary Driving Simulator'', proceedings of the IEEE International Conference on Systems, Man and Cybernetics, vol 2, pp 711-1716, 1995 Developing Electronic Power Steering and Steer-byWire Systems”, proceedings of the IEEE/RSJ International Conference on Intelligent Robots and Systems, pp 1374-1379, Piscataway, NJ, 1999 [13] Gillespie, B R., Hasser, C., and Tang, P., "Cancellation of Feed Through Dynamics Using a Force-Reflecting Joystick'', proceedings of the ASME IMECE, Nashville, TN, 1999 [14] Setlur, P., Dawson, D., Chen, J., and Wagner, J., "A Nonlinear Tracking Controller for a Haptic Interface Steer-by-Wire Systems", proc IEEE Conference on Decision and Control, Las Vegas, NV, 2002 [15] Lewis, F L., Abdallah, C T., and Dawson, D M., Control of Robot Manipulators, Macmillan Publishing Company, 1993, ISBN 0-02-370501-9 [16] Tilin, A., "Redesigning the Driver - You Are About to Crash", Wired, vol 10, no 4, April 2002 [17] Wagner, J., and Keane, J., "A Strategy to Verify Chassis Controller Software - Dynamics, Hardware, Automation", IEEE Transactions on Systems, Man, and Cybernetics, vol 27, no 4, pp 480-493, 1997 [18] Andonian, B., Rauch, W., and Bhise, V., "Driver Steering Performance Using Joystick vs Steering Wheel Control", SAE paper no 2003-01-0118, 2003 Mission Force/Torque Input Human Operator Graphical Display Visual Information Force Feedback Audio Wheel or Joystick Interface Directional Control Assembly Environment Sound SCU Real Time Virtual Environment Vehicle Dynamics Figure 7: Steer-by-wire test station with virtual reality, cockpit, and input mechanisms 2589 Proceedings of the American Control Conference Denver, Colorado June 4-6, 2003 ... experimental testing A real time hardware-in-theloop (e.g., [17]) experimental test platform has been constructed to provide a virtual reality interface for operators to drive the "vehicle" and evaluate... force feedback and/ or steer the vehicle Additional active and passive devices have been added at the rack to apply different road reaction forces A second computer workstation performs virtual. .. Evaluation of On-Center Handling", SAE Paper No 960178, 1996 [10] Setlur, P., Dawson, D., Wagner, J., and Fang, Y., Milsap, S., and Law, E., “Handling Enhancement due to an Automotive Variable