Combined braking and steering maneuver for collision avoidance

Abstract

Abstract In modern vehicles today, safety features such as Automatic Emergency Braking (AEB) and Automatic Emergency Steering (AES) are important for collision avoidance and reducing traffic fatalities. AEB and AES are constrained by their inability to handle complex scenarios. Therefore, further development is required to explore new ways to avoid collisions in complex scenarios. This thesis proposes two algorithms, Rule-based and Spline-based. Both utilize combined braking and steering with multi-target threat assessment for collision avoidance. The Rule-based algorithm relies on multi-target threat assessment and free space calculations, to determine when to steer or brake. The Spline-based algorithm samples several quintic spline trajectories every time step. If the algorithm finds collision-free, dynamical feasible trajectories, the safest trajectory is selected with a cost function. Both algorithms manage to generate collision-free trajectories and prevent collisions if the trajectory is dynamically feasible. The two algorithms in some test scenarios choose different combinations of braking and steering, to avoid collision.