Path Planning of Industrial Wheeled Robots Based on Wireless Communication and Machine Learning Algorithms

Abstract

With the advancement of science and technology, robotics has made considerable progress. Robots can free humans from heavy repetitive labor. From the industrial field to the lives of the general public, robots are playing an increasingly important role. Path planning is one of the core contents of industrial wheeled robotics and has very important significance. Based on the reinforcement Q learning and BP network, this work studies the path planning of industrial wheeled robots. According to task requirements of path planning, design learning strategies, and control rules, use wireless communication to transmit environmental perception information and propose corresponding control strategies. The main researches are as follows: (1) Based on grid map environment, a path planning algorithm with Q-CM learning is designed. The algorithm first designs robot states and actions based on reinforcement Q learning and grid map and establishes Q matrix. Secondly, a coordinate matching (CM) obstacle avoidance control rule is designed to improve the efficiency of robot avoidance. Then, a reward function is designed for the evaluation problem of action execution. (2) Based on the map environment of free space and the generalization ability of BP neural network, a robot path planning with Q-BP learning is designed. The algorithm first designs the sensor detection mechanism and action selection strategy according to the state of the robot in the map environment. Secondly, a dynamic reward function is designed. Then, according to the obstacle avoidance requirements of special obstacles, the obstacle avoidance rules after three shocks were designed. The experimental results show robots can perform better path planning in a discrete and continuous free space map, and the obstacle avoidance effect is good.