An Algorithm Based on Neural Network for Mobile Robot Path Planning

Abstract

Path Planning technology for mobile robots is an important subject of robotics studies.For unknown global environment,in this paper,the whole task is decomposed to a series of sub - tasks with known and simple environment through the robot's sensor network.Then a discrete neural network penalty function is constructed by using the parallel computing predominance of neutral network,and then a variable - parameter method is introduced to obtain the negative gradient direction of the penalty function and solve the sub-tasks fast and efficiently.So,the robot can be driven to the goal.This strategy simplifies the path - planning algorithm in complex environment,and improves the control effects.Especially for a real - time situation,it has distinct advantages.