Multi-robot motion planning method and system and storage medium

Abstract

The invention provides a multi-robot motion planning method and system and a storage medium. The method comprises the following steps: acquiring the motion state and environment information of robots in motion; determining a motion strategy of the robot through reinforcement learning according to the discretized motion state and environment information, wherein the continuous action in the motion strategy is obtained by determining a reinforcement learning state space by using a fuzzy neural network and outputting the obtained continuous action; determining a basic behavior of the robot according to the motion strategy, and performing cluster motion, wherein the basic behaviors comprise at least one of the following behaviors: advancing to a target, obstacle avoidance movement, collision avoidance movement and movement along a wall. The method solves the problems that the convergence speed is too low and the predictive ability is poor when optimal behavior strategy learning is carried out on the intelligent body in a huge state space and a dynamic change environment, and can be widely applied to the technical field of robot control.

Description

technical field [0001] The invention relates to the technical field of robot control, in particular to a method, system and storage medium for multi-robot motion planning. Background technique [0002] Motion planning is a hot topic in the research of multi-mobile robots. At present, the motion planning of multi-mobile robots mainly includes three control behaviors: path planning, formation control, and obstacle avoidance and collision avoidance. Among the many motion coordination algorithms, a new coordination method - flocking (Flocking) control mode is a new type of distributed control method that simulates the biological aggregation movement in nature. The three aspects of obstacles and movement towards the goal point coincide exactly with the three models of swarm movement: separation, alignment and convergence. [0003] The advantage of cluster control lies in formation gathering, stable formation, and the ability to avoid certain obstacles and move towards the targe...

AI Technical Summary

Problems solved by technology

However, the traditional reinforcement learning used in the autonomous robot motion planning system in an unknown dynamic multi-obstacle environment will face three problems: 1) when the state space and action space are continuous or the number is too large, the reinforcement learning convergence speed is too slow; 2) the agent is learning In the early stage, it is a blind search, and the foresight ability is poor; 3) The reward function value of the environment for each action cannot be clearly defined

Images