An Interactive Game Theory-PSO Based Comprehensive Framework for Autonomous Vehicle Decision Making and Trajectory Planning

Nouhed Naidja (1, 2) , Stéphane Font (1) , Marc Revilloud , Guillaume Sandou (1)

1 L2S - Laboratoire des signaux et systèmes
2 VEDECOM - VEDECOM

The mutual dependence between autonomous vehicles and human drivers is an open problem for the safety and feasibility of autonomous driving. This paper introduces a game-theoretic trajectory planner and decision-maker for mixed-traffic environments. Our solution accounts for interaction with the surrounding vehicles while making decisions, and uses a clothoid interpolation method to generate human-like trajectories. The Particle Swarm Optimizer (PSO) used here bridges the decision-making and the trajectory generating processes for a joined execution. We chose an unsignalized intersection crossing scenarios to demonstrate the feasibility of our method. Testing results show that our approach reduces the dimension of the search space for the trajectory optimization problem and enforces geometric constraints on path curvature

Keywords: Decision Making; Game Theory; Trajectory optimization; Particle Swarm Optimization

Domains: Engineering Sciences [physics]

Nouhed Naidja, Stéphane Font, Marc Revilloud, Guillaume Sandou. An Interactive Game Theory-PSO Based Comprehensive Framework for Autonomous Vehicle Decision Making and Trajectory Planning. IFAC World Congress - 22nd WC 2023, International Federation of Automatic Control, Jul 2023, Yokohama, Japan. ⟨hal-04102621⟩