Improving heavy duty vehicle stability during split-μ braking situations with rear axle steering

Sammanfattning

This thesis investigates the use of rear axle steering (RAS) to improve the stabilityof heavy-duty vehicles during split-μ braking scenarios. Split-μ braking, where thefriction coefficient differs between the left and right wheel paths, generates asymmetriclongitudinal forces and induces yaw moments that can compromise vehiclestability, control and safety.Controllers for the RAS actuator were developed and evaluated, including PID control,gain-scheduled Linear Quadratic Regulator (LQR), robust H∞ control, adaptiveModel Predictive Control (MPC), and a brake pressure feed-forward approach.The controllers were implemented and assessed in a co-simulation environment usingMATLAB/Simulink and IPG TruckMaker. The stability properties of the controllerswere analysed using an induced-norm-based metric to assess their influenceon the driver’s intended vehicle behaviour. Performance was evaluated across multiplescenarios, including straight-line braking, curved-road braking, and lane-changemanoeuvres under various slit-μ conditions. Performance metrics included yaw rate,yaw angle, lateral deviation, braking distance, and driver steering effort.The results demonstrate that RAS can significantly improve lateral stability andreduce driver effort without notably increasing braking distance. Among the evaluatedmethods, the LQR and PID controllers provided the best overall performancein terms of stability and driver workload reduction, with the LQR offering a balancebetween performance and control effort. The H∞ controller showed robustness, whilethe MPC showed less consistent performance for short-duration, high-dynamics manoeuvres.The feed-forward approach proved effective in reducing initial yaw disturbanceswhen combined with feedback control. Furthermore, it is shown that a moreaggressive ABS strategy can be applied without compromising stability when combinedwith RAS control, thereby improving both braking performance and stability.The results also show that performance is primarily limited by the steering rateof the RAS actuator rather than by its maximum angle. Additionally, the LinearParametric Varying (LPV) reference model used for control design can be furtherextended to improve controller performance and expand the operational capabilityof the RAS system.Overall, the thesis confirms that rear axle steering is a viable approach for enhancingvehicle stability during critical split-μ braking scenarios and can be implementedusing existing vehicle signals without requiring predictive sensing technologies.