Motion planning for an AGV fleet in a logistics scenario

Abstract: As warehouses are becoming more autonomous the number of Autonomous Guided Vehicles (AGV) increases. Ensuring that the motion planning is safe despite the large number of agents and despite sharing the workspace with humans is the problem we tackle in this thesis. We propose a solution based on velocity obstacles that ensures the collision avoidance and a global planner that takes into account the human comfort. This solution is evaluated in simulation with dynamic unicycle type robots with and without inter-robot communication and offers improvements over Hybrid Reciprocal Velocity Obstacles (HRVO) in the case without communication.