Optimal control of brakes and steering for autonomous collision avoidance

Paper i proceeding, 2018

This paper deals with collision avoidance for road vehicles when operating at the limits of available friction. We separate the control problem into two phases: collision avoidance phase and lane keeping phase. The aim is to reduce the amount of overshoot after avoiding the collision. The mass-center acceleration reference is obtained from a particle motion and implemented via a novel lower-level control allocation method, the Modified Hamiltonian Algorithm (MHA). At the upper level, we compared 3 common collision avoidance strategies with the proposed control strategy and determine trade-offs between the available intervention distance vs. overshoot during the lane keeping phase. It is found that the proposed control strategy has the least overshoot. MHA is found to achieve high quality tracking of the particle acceleration and it has performance close to that of the ideal particle. Yaw moment control, an integral part of MHA, is also found to be effective in maintaining yaw stability throughout.