Complete Coverage Path Planning Based on Bioinspired Neural Network and Pedestrian Location Prediction

Abstract

Mobile robots' complete coverage path planning (CCPP)based on bioinspired neural network can deal with real-time coverage task in dynamic environment efficiently. But in most of the existing results, passive obstacle avoidance strategies are used, which may lead to the “collision problem” between robot and pedestrian or other moving obstacle. This paper will propose an improved CCPP approach with “active obstacle avoidance strategy” based on bioinspired neural network and pedestrian location prediction. In the proposed approach, the trajectory of the moving pedestrian is firstly predicted by Kalman filter algorithm. Then the predicted location is introduced to the general neural network, changing the external inputs for activity landscape. Furthermore, an optimal avoidance strategy is designed for mobile robots. Finally, a pedestrian location prediction based CCPP approach is presented, which integrates biologically inspired neural network, location prediction algorithm and pedestrian avoidance strategy. Simulation results show that the proposed is feasible for the robot to plan a collision-free trajectory of CCPP.