THE CFD-BASED NEURAL DYNAMIC PATH PLANNER FOR MULTI-ROBOT SYSTEM

Abstract

Real-time path planning is a crucial research issue for multi-robot systems (MRSs), particularly in a 3-D environment with obstacles. In this paper, combining with computational fluid dynamics (CFD), an improved neural dynamic path planner is proposed for MRS. It is capable of real-time planning the paths while avoiding obstacles for MRS in complex, dynamic and large-scale environments. The proposed path planner has lower computational complexity than traditional ones. The computational complexity of the proposed path planner is not sensitive to the number of individuals in MRS. The validity of the proposed approach is demonstrated by comparative simulations with different simulation scenarios.