Trajectory Optimization for Autonomous Vehicles on Crossroads with Mobile Obstacles
Jean-Baptiste Receveur, Stéphane Victor, Pierre Melchior · 2018
Autonomous vehicles issues have emerged in the past years. Among the classic trajectory planning methods, the potential field method, which is mainly local, is purely reactive to the environment and creates rough trajectories. In this article the creation of a dynamic optimal minimum of the potential is proposed, using multi-criteria trajectory optimization, for unmanned terrestrial vehicles. Obstacles are mobile and with the anticipation of their movement, motion and path of this optimal dynamic minimum are jointly generated. The effects of adding this anticipation in the planning are illustrated, and the optimization is done using a Genetic Algorithm (GA), to improve the Potential Fields (PF) method. In the first two parts of this article, the problem is formulated and the different criteria are described. Then, a GA-PF method is proposed. Finally simulations of the potential field method and its combination with a genetic algorithm are presented.