Fig. 1.1 Concept cars equipped with a steer-wire-system

Transcription

1 THE REALISTIC HAPTIC FEEDBACK OF STEERBYWIRE SYSTEMS BASED ON A DIRECT CURRENT MEASUREMENT METHOD Ba Hai Nguyen 1 and JeeHwan Ryu 1 1 School of Mechanical Engineering, Korea University of Technology and Education, Cheonan, Korea (Tel : ; bmnhy2003@gmail.com) 2 School of Mechanical Engineering, Korea University of Technology and Education, Cheonan, Korea (Tel : ; jhryu@kut.ac.kr ) ABSTRACT: A novel solution for implementing the realistic haptic feedback of steerbywire systems is shown in this paper. First, requirements of steerbywire systems and current solutions are surveyed and discussed. Second, we introduce a direct current measurement method, which is used to artificially create haptic feedback, and a new algorithm for handling free control of the handwheel () in such systems. Finally, simulation results have proved that the realistic haptic feedback and maneuverability as well as stability of the can be abstained with based on the proposed method. In addition, the direct current signals are also benefit for conducting a redundant feature and diagnostic information for steerbywire systems. The method mentioned in this paper offers a cheap solution and a simple algorithm for implementation of steerbywire systems. Keyword: steerbywire, realistic haptic feedback, free control, humanmachine interface. 1 INTRODUCTION A steerbywire technology promises a number of benefits which are the weight reduction of mechanical systems and increasing free space in the cabin, as well as enabling advanced safety functionalities such as vehicle stability control and lanekeeping assistance [1]. Besides that, completely replacing conventional steering systems with steerbywire systems make the possibility for removing the steering shaft, column and gear mechanism. Therefore, a new opportunity which gives designers better chances for car interior design. In addition, the fuel consumption could be achieved due to the lack of hydraulic pumps. Without mechanical connection between the and the road wheel, human injury could be reduced in case of car crashes. However, until 2008, except some concept cars like Hywire, Filo, there is no production car available that installed with a steerbywire system due to safety and reliability concerns. And, since the safety of steerbywire system is strongly related to the driving feel known as realistic feedback [1],[2]. This is a critical factor for giving drivers perception of driving condition and vehicle states as well as the road properties. However, this driving feel is removed physically due to the lack of the mechanical linkages in the system. SKF Filo Nissan Pivo BWM Concept car GM Hywire Fig. 1.1 Concept cars equipped with a steerwiresystem Recently, several solutions for overcoming this drawback have been given by several researchers worldwide [1],[3],[4],[5][6]. A Torque map method is introduced by SeWook Oh & et.al., from Hyundai

2 Mobis, in 2006 [3] with a position control for the roadwheel (RW) subsystem and a force control loop for force feedback of subsystem. Model based approach is proposed by professor, J. Christian Gerdes at the dynamic design laboratory, Stanford university in 2004 [4]. Haptic feedback based on torque sensors is the idea which comes from the field of robotics proposed by PI technology, UK [9]. However, the need of a cheaper and simpler solution for haptic feedback and steerbywire control is still a challenge for automotive manufactures and researchers today. This is also the main motivation and the content of this paper. Many related issues will be discussed including the realistic haptic feedback implantation for, free control stability and steerbywire redundancy. This paper is organized in six sections. In the first section, an introduction of the research motivation and objectives including declaration of the open problems related to steerbywire development and current solutions are presented. The basic knowledge of the steerbywire system and requirements of a steerbywire system are provided. The detail advantages and the disadvantages of current solutions will be discussed deeply in the second section. The core of this paper, which is the direct current measurement method, will be comprehensively showed in the section three. In the fourth section, the proposed method is investigated with the simulation results. The summary of research and future works are given in the last section. 2 REQUIREMENTS OF SBW SYSTEM AND CURRENT SOLUTIONS 2.1 Requirements: The Fig. 2.1 displays the steerbywire conversion from a conventional steering system. The key concept to make a steerbywire system is to replace mechanical/hydraulic linkages with electronic components such as encoders, motors and microcontrollers. Due to this lack of mechanical connection, it is clear to realize that a steerbywire system must to fill up with all of the following requirements and constraints. actuators upper intermediate shart lower joint power device controller RW sensors RW actuators sensors rack RW Fig. 2.1 Conversion from conventional steering system to SBW Position tracking, this is the fundamental function of any steering systems. In conventional steering systems, the RW position is set by a driver thought the angle input and mechanical/hydraulic components. Thus, for steerbywire system, this position will be electronically transmitted through a electronic connection governed by the combination of the sensors and the controllers. For this reason, position sensors are attached to the and the RWs' vertical axes. Realistic haptic feedback, which give driving feel as conventional driving feel in the previous versions of steering system. However, the most difficult questions for all steerbywire systems are what feedback signal can be used? How much is the force feedback for the system? And how will the force feedback be implemented? And of course, how much will it cost to bring the realistic haptic feedback to the commercial vehicles? Free control, which refers to the response of the after a certain condition given to the wheel by the driver. A quick return to center with minimal overshoot is desired [5] as free control is designed. Furthermore, underdamped or sustained oscillations about the center are highly undesirable. Basically,

3 once the returning forces from hydraulic, EPS, etc react to steering system; the response of free control is occurred. In the case of bywire, the poor response starts once the driver removes his hands. Therefore, free control is essential for the steerbywire system. Redundancy, if the throttle, ignition and transmission and steering system are all beyond the direct control of the driver there is no effective ways for handling the steering system and the vehicle in case of any faults occur. Therefore, faulttolerance and redundancy are the critical issues for bywire systems. 2.2 CURRENT SOLUTIONS: Torque map The control structure of this method is represented as Fig There are two control loops. The first loop is force PID control. The feedback device used for giving the actual torque value of is a torque sensor. The second loop is the position control loop for tracking the input signal of given by the driver. Driver s steering torque Steering column angle factor Torque sensor ECU motor torque Displacement sensor Gear ratio RW angle Rack assist motor Feed forward control Fig. 2.2 Control scheme of the torque map method A reference driver's steering toque is preprogrammed based on the reactive force feedback of the angle effect and the velocity effect. Those two effects can be changed by the adjustment of each constant gain. The detail effect of each term is represented in the following equations: y SW = θ (2.1) yv = Kβ x ( x VV max) Tin (2.2) 3 2 where y : reactive force related to the angle SW y V : reactive force related to the vehicle s velocity x : vehicle s velocity (this denotation is used only in torque map method) max : maximum vehicles speed, Tin is the initial torque V V K α Model based approach This solution is proposed by professor, J. Christian Gerdes in However, the disadvantage found here is that the vehicle dynamics and a perfect dynamic model of the subsystems must be known

4 for calculating the lateral forces Fig Also, the limitation of automotive ECUs' capacity is a significant problem in this case [3]. 2 st order 1 st order Js bs feedforward Ka Λ τa 1 st order 1 st order sg n Fs Kp Kd friction feedback τ τaligning Steering system θ α f Vehicle dynamics θ 1 st order Fig. 2.3 The control scheme of the modelbased approach Torque sensor In this solution, huge torque sensors are mounted on the rack at the joint of the rack and the steering arm [9]. This method can easily feedback all the effect which come from inertia, damping of the system as well as the aligning moment caused by the lateral forces at the RW. However, torque sensors are very expensive. Attaching some torque sensors in the system It is also not suitable due to the severe working condition of the vehicles including dusts, oil, high temperature, and vibrations. As we mentioned in this section, there are several solutions, and methods which have been proposed for solving the realistic haptic feedback issue as well as for the general steerbywire development. Each method has different disadvantages. The essential motivation for our research is from those. We have done a lot of efforts to simplify the complexdynamics of the system and the cost of torque sensors by using direct current measurement method which is going to be discussed in the next section. 3 DIRECT CURRENT MEASUREMENT METHOD 3.1 System overview In our proposed system, there are two control loops. An encoder detects movements like direction, magnitude, speed; and sends inputs to an onboard controller. It interprets the signals and then commands electromechanical actuators to turn the wheels accordingly. Position sensors on the actuators close the feedback loop with the controller, ensuring the vehicle wheels match steering commands Fig The current sensor unit is connected to RW's motor driver and RW's motor in series for measuring current of the motor. The vehicle speed and the angle are needed for creating the assistance force due to the lack of hydraulic pump. Free control algorithm is involved to the haptic feedback generator. This free control technique is discussed deeper later.

5 PID Position Controller Motor Driver Motor Driver Force Controller Current Sensor Unit Free Control angle Assistance Torque Generator Vehicle speed & angle Fig. 3.1 System overview 3.2 Haptic feedback implementation To find how the haptic feedback should be implemented, the model of conventional steering system shown in Equ. 3.1 would be considered. This is necessary and useful to know how to modify and artificially create the realistic haptic feedback and to design the free control algorithm later. J wδ b w δ τ f τa = τm (3.1) where: J, : are inertia constant, and damping constant of the steering system respectively w b w ** * δ,δ τ τ, τ f a M : are RW angular acceleration, and RW angular acceleration respectively, : are Coulomb friction, and aligning toque, and torque of RW actuator respectively Now, the direct measurement method for the development of steerbywire systems can be explained. First of all, position tracking is done by the PID control loop. Two encoders are used. The first one is to determine the position commanded by the driver. As the PID position control works, the RW's actuator is activated to get the motor running until the error between the motor and RW motor reaches zero. When the RW motor works (to complete the position control loop), the motor current increases or decreases depending on the load of RW motor. The error here strongly relates to the dynamics of the steering system, RW friction, and mechanical geometry of the steering system, vehicle states or even the road condition and so on. Second, the current signal is used to feedback to the force feedback controller. This is the second loop in the steerbywire control. This loop includes the assistance force generator which composed by the effect of angle (include frequency of angle), and the vehicle velocity. The current sensor unit is connected to RW's motor driver and RW's motor in series. This connection allows us to measure the current runs inside the RW actuator. All the dynamic behaviors of the system will affect on the motor shaft, and then this dynamic property is translated in current signal. In other words, the current sensor unit will be observed all the changing dynamics related to the steering

6 system including the aligning moment. By doing this, there is no need for the knowledge of the vehicle dynamics and steering dynamics which are the most important for the modelbased approach discussed in the previous section. and therefore, the haptic feedback which is needed to transmit to the driver is calculated from the Equ τ = G ( K. i τ ) T (3.2) feedback where: τ : is the haptic feedback for the driver feedback G feel : K t, i : τ assist : T : is the feeling gain feel t assist are the motor s torque constant and motor current respectively is the assistance torque is a free control torque. The free control torque is shown detail in chapter six In order to reduce the force feedback at the lower speed (the speed of the vehicle) or tighter at the higher speed, the force feedback will be reduced by assistant toque τ assist. τ assist 2 K θ signθ ) KV = (3.3) K, K a v : are the angle and velocity gain constants respectively V, v T : are the vehicle s speed and free control torque respectively. a ( * v v 3.3 Free control Traditionally, free control response has been resolved by adding damping (in the case of EPS) or friction (in the case of hydraulic steering) to the system [5]. Our proposed idea is to add additional damping to steerbywire system. Thus, we now introduce a new torque to the system called free control torque. This torque is a product of a free control gain K, and the angle. By introducing this torque to the system, the force feedback will change. Fortunately, because the special design and property of steerbywires allow us to quickly & easily adjust the force feedback by changing the angle and velocity gains or the scaling factor of RW motor's torque. The free control torque can be expressed under this equation: T K *θ = (3.4) 4 SIMULATION RESULT Position tracking, a good position tracking also can be achieved in this method shown in Fig The simulation parameters are: Kp=5.5; Ki=0.008; Step size: (s); The result shows that error of the system is very small (about error: rad). This is sufficient comparing to the normal error of steerbywire system (about rad)

7 Fig. 4.1 Good position tracking of set RW angle vs. actual angle Realistic haptic feeback, Fig 4.2 and Fig. 43 are the comparison of simulation of torque map method and the proposed method. The simulation result clearly shows that, the proposed method can achieve the realistic haptic feedback which provides the driving feeling for intuitive steering control. Fig. 4.2 position vs. Force feedback in torque map method Fig. 4.3 position vs. Force feedback in torque map method In addition, with the direct current measurement method, the force feedback can be abstained and easily adjusted the maximum torque by changing the feeling gain as well as the tradeoff of the assistant torque gain and the free control gain. Free control, the simulation result in Fig. 4.4 and Fig. 4.5 has proved that, the free control torque has significantly reduced the oscillation of. This could be explained easily because when the free control factor is added to the system, the hysteresis of steering system is reduced shown in Fig The hysteresis of proposed method is reduced to 5 to 5 degrees. It was from 10 to 10 degrees in torque map method shown in Fig Fig. 4.4 The behavior without free control torque ( T ) (Without scaling of angle and WR angle) Fig. 4.5 The behavior with free control torque ( T ) (Without scaling of angle and RW angle) 5 CONCLUSION A new solution was proposed for improving the performance and safety of steering systems. The proposed method could be a cheap and easy to implement with simple algorithms. The problem of complex dynamics was solved based on the direct current measurement method. The simulation results clearly proved that the realistic haptic feedback for steerbywire system can be obtained by using this proposed method. Free control and maneuverability also were improved by reducing returnability hyteresis. In addition, the measurement signal could be used to realize the faults of SBW systems without mechanical/hydraulic devices. And, this would be recommended to complete in the near future including

8 the study of frequency effect on the steerbywire systems, the Kubiro xbywire vehicle (KUTBiorobotics xbywire vehicle) development, and the improvement the torque map for SBW controllers, and the study on stability of the steerbywire system. REFERENCE [1] Paul Yih. Steerbywire: Implications for Vehicle Handling and Safety January 2005 [2] Sanket Amberkar, Farhad Bolourchi, Jon Demerly and Scott Millsap, A control system methodology for steer by wire systems, Steering and Suspension Technology Symposium, SAE technical paper series, [3] SeWook Oh and e.t. The Design of a Controller for The SteerByWire System, JSME International Journal, [4] Yih, P. Ryu, j. Gerdes, J.C. Vehicle State Estimation Using Steering Torque, American Control Conference, july [5] Douglas Cesiel and Michael C. Gaunt, Development of a SteerByWire System for the GM Sequel, General Motors Corporation. [6] F. Bolourchi & C. Etienne; Active Damping Controls Algorithm for an Electric Power Steering Application, Proceedings of: 30th International Symposium on Automotive Technology & Automation pp ; June 97. [7] Sanket Amberkar, Farhad Bolourchi, Jon Demerly and Scott Millsap, A control system methodology for steer by wire systems, Steering and Suspension Technology Symposium, SAE technical paper series, [8] Dr. Juan R. Pimentel Kettering University, Flint, MI, USA An Architecture for a SafetyCritical SteerbyWire System, 2004 Society of Automotive Engineers, Inc. [9] JB Drive By Wire Pininfarina Autosicura, 2004 PI Technology.