Sliding mode techniques for automotive vehicle dynamics

Numerous control techniques, including full state-feedback sliding mode control, have addressed the split-m braking manoeuvre. The controllers presented in this thesis extend previous work by using only certain measured outputs. These controllers work in conjunction with an anti-lock braking system (ABS) to provide safe, effective braking through steer-by-wire. Two strategies are presented, each using two measured outputs (yaw rate and lateral deviation) and front road wheel steering angle as the sole input. The first scheme estimates two further states by employing a sliding mode observer. It benefits from sliding mode robustness properties, and is demonstrated by simulation on a nonlinear model to be robust to large variations in tyre stiffness. The second scheme consists of a compensator-based, sliding mode controller. This controller, tested under the same conditions as the first, does not perform as well as the observer-based controller from a robustness perspective, but it produces a closed-loop system of lower order, which may be advantageous from an implementation perspective. Potentially dangerous vehicle scenarios, such as severe understeer, oversteer and split-m braking, are currently detected using a large combination of measurements, seeking to estimate vehicle states robustly. In this thesis, a simpler approach is adopted, which, rather than estimating all the vehicle states, looks for `signature differences' between the behaviour of an ideal linear vehicle model and the actual measured behaviour, in terms of vehicle yaw moments and lateral forces.