Vehicle Side Slip Angle Observation with Road Friction Adaptation

Abstract

Side-slip angle is an important variable indispensable for some advanced automotive vehicle control systems. However, due to technical and economic reasons, this variable is often estimated rather than directly measured. In this paper we address the observer design problem for estimating the side-slip angle and the unknown road friction coefficient, based on measured signals from sensors common to modern series-production automobiles. We formulate state and parameter estimation as a non-convex optimization problem. By interweaving discrete time solution of the optimization and continuous integration of sensor data, our scheme allows for sufficient time for finding the global optima approximately through a grid-search. Consequently, despite the non-convex optimization we are facing, our observation scheme is able to run in realtime. We show some desirable properties of the proposed scheme concerning the stability and convergence of estimation error. One advantage of our observer is that for the nominal model the estimation error does not grow even when the system lacks observability. Simulation shows that the proposed observer provides very accurate estimation in challenging scenarios where the vehicle executes extreme maneuvers and measured signals are corrupted by noise.