This article presents an innovative optimization-based solution to the collision avoidance challenge for autonomous vehicles. The presented approach consists in an online motion planner designed to define feasible and efficient paths able to deal with dynamic surroundings while implicitly ensure safety in the proposed maneuvers. The fact of considering moving obstacles inside the motion planner increases the complexity of the problem while forces it to be executed more frequently as others. To reduce this computational complexity, the approach presented counts with a two stages translation of the commonly used non-linear optimization-based structure into a QP formulation which can be easily solved. The first stage is based on the use of LPV matrices in the dynamic constraints of the vehicle. The second stage consists in performing a reachability analysis based on set propagation to obtain linear expressions of the permitted inputs and reachable states which guarantee safety conditions.

This work has been co-financed by the Spanish State Research Agency (AEI) and the European Regional Development Fund (ERFD) through the project SaCoAV (ref. MINECO PID2020-114244RB-I00) and by the DGR of Generalitat de Catalunya, Spain (SAC group ref. 2017/SGR/482).

Peer Reviewed