Multipath Planning Based on Neural Network Optimized with Adaptive Niche in Unknown Environment

Abstract

An algorithm for multipath planning in unknown environment is presented. Niche identification technology used in multimodal function problems is improved and is applied to path planning. Fitness sharing method is adopted to keep diversity of niches. The robot detects local environmental information with sensors and its movement is controlled by neural network. The neural network is optimized by the genetic algorithm based on adaptive niche. There is no need for generating feasible paths at first and clustering in every generation. The simulation results show that the algorithm is efficient.