Decentralized Receding Horizon Motion Planner for Multi-robot with Risk Management

This paper proposes a decentralized motion planner for a fleet of autonomous mobile robots in the presence of hazardous risk situations. First, some risk cases related to hazardous industrial facilities are identified and discussed. Safety requirements associated with the identified risk situations are defined in terms of constraints that must be considered in the trajectory planning problem. For instance, once a robot enters a hazardous area, the remaining robots consider such a hazardous area as a virtual obstacle that must be avoided. Based on these constraints, a receding horizon motion planner is introduced. To provide a fully decentralized scheme, a three-step sequence is provided where each robot presumes a trajectory for other robots belonging to its conflict set. To solve the planning problem, B-spline parametrization and a particle swarm optimization algorithm are used. A numerical simulation has been conducted to show the feasibility of the proposed scheme.