Multi-agent Poli-RRT* Optimal constrained RRT-based planning for multiple vehicles with feedback linearisable dynamics

Planning a trajectory that is optimal according to some performance criterion, collision-free, and feasible with respect to dynamic and actuation constraints is a key functionality of an autonomous vehicle. Poli-RRT* is a sample-based planning algorithm that serves this purpose for a single vehicle with feedback linearisable dynamics. This paper extends Poli-RRT* to a multi-agent cooperative setting where multiple vehicles share the same environment and need to avoid each other besides some static obstacles.