An improved neural dynamics based approach with territorial mechanism to online path planning of multi-robot systems

摘要

The coordination of multi-robot system (MRS) are applied commonly to various fields of the automotive industry. In a variety of cooperative modes, online path planning with obstacles avoidance is a fundamentally important hotspot, especially in a 3-D, complex, or dynamic environment. In the paper, an improved neural dynamics based approach with territorial mechanism is proposed to online path planning of MRS, which can be used as the online path planner for multi-AUVs and multi-UAVs in complex and dynamic environments. This approach integrates biological neural network, computational fluid dynamics, and territorial mechanism of animals, which has the characteristics and advantages of the biological nervous system, namely self-regulation, self-adaptation, self-organization, etc. It can cope with a variety of accidents during path planning, such as the disappearance of targets, the breakdown of robots, the change of environments, and so forth. Meanwhile, the proposed approach has better time performance and is insensitive to the number of robots in MRS. During the path planning of MRS, it can also guarantee to balance workload and to reduce entire workload and total time, which enhance robustness and fairness. The effectiveness and efficiency of the proposed approach are demonstrated by simulations and comparative studies.