A Path Planning Algorithm for Sweeping Robot Based on Improved Neural Network

Abstract

This paper plans to solve the problem of the lacks of traditional path planning algorithm in self-learning and the disadvantages in local path planning. We propose a path-planning algorithm for sweeping robots based on improved neural network. It is aimed to provide a collision-free planning route for robots in an unknown environment. This algorithm is based on the neural network algorithm, which concludes a four-layer structure of network, and designs the energy function as the evaluation function of the network. First I need to find the extreme value of the energy function, and then I need to adjust the motion based on the path points, next, the robot completes the path planning task. The effectiveness of the method is proved by simulation experiments.