Using Mems Accelerometers to Improve Automobile Handwheel State Estimation for Force Feedback

This paper explores the possibility of combining solid state accelerometers with a low resolution position sensor to provide clean estimates of automobile handwheel position, velocity and acceleration for use in force feedback. Typically determining the acceleration and velocity of the handwheel requires differentiating a position sensor such as an encoder or potentiometer. Unless expensive high-resolution sensors are used, this differentiation leads to a noisy signal, requiring significant filtering which leads to significant phase lag. With a direct measurement of acceleration, we circumvent many of the problems associated with differentiation and filtering. This work uses a Kalman filter to combine a pair of MEMS accelerometers with a low-resolution potentiometer to estimate handwheel states. This measurement scheme is effective in this application because of the low frequency nature of the force feedback and because of the structural robustness of the handwheel system. Initial in-vehicle experimental results show the setup can provide smooth acceleration and velocity signals in a moving vehicle. NOMENCLATURE δ Handwheel angle b Total accelerometer bias Ad Discrete state transition matrix Bd Discrete input matrix Cd Discrete output matrix Address all correspondence to this author. Lss Steady state Kalman gain Σss Steady state error covariance Σv Sensor noise covariance Σw Process noise covariance x Estimate state vector INTRODUCTION The handwheel is the primary control mechanism of an automobile, and thus interaction between the handwheel and the driver is critical to safety. Forces on the handwheel communicate to the driver about the tire forces acting on the vehicle. The handwheel also presents a predictable mechanical feel to the driver to allow smooth and safe control. In a vehicle with a conventional steering system, forces are transmitted from the road wheels to the handwheel through the mechanical steering linkage. These include forces from road imperfections and the aligning moment. The aligning moment is what causes the steering wheel to return to center while the vehicle is moving. It is caused by the offset between the steering axis and the application point of the tire side force in the contact patch (see Figure 1). This offset is a combination of the mechanical and pneumatic trails. The pneumatic trail is the offset between the center of the tire patch and the centroid of the side force generated by the tire. This is a result of the mechanism by which tires generate lateral force. The mechanical trail is a geometric offset caused by the design of the steering geometry. The effective lever arm is the sum of these two offsets, and the tire 1 Copyright c © 2004 by ASME Mechanical Trail Steering Axis Pneumatic Trail Centroid of Tire Force Figure 1. STEERING GEOMETRY CAUSING ALIGNING MOMENT. force applied about this arm causes a moment about the steering axis that tends to straighten the wheels. In a conventionally steered vehicle, in addition to acting to straighten the wheels, this moment also provides the driver an indication of the forces on the road wheels, and can warn of an impending loss of traction. Steer-by-wire vehicles require artificial force feedback to replicate these forces in a conventional steering system. Steerby-wire differs from conventional steering in the way the handwheel controls the roadwheels. In a conventionally steered vehicle, the handwheel is mechanically connected to the road wheels, and the driver controls the steering system of the vehicle through this mechanical linkage. In a steer-by-wire vehicle this mechanical link has been removed, and some sort of actuator controls the road wheels in response to commands from the handwheel (see Figure 2). Steer-by-wire systems of the future will increase vehicle safety, simplify vehicle cabin design, and open up new possibilities for vehicle control. With the mechanical steering link removed, any force feedback at the handwheel must be artificially created using a motor or other actuator. This includes the aligning moment, but must also include an appropriate amount of damping and inertia to