Method of Real-time Environment Learning and Robot Navigation Based on Distributed Perception-behavior Network

A distributed perception-behavior network is used to represent environment for robot,which lacks a priori environmental model,global localization information and compass information,from its own spatio-temporal experiences.because the model in distributed,the robot can localize in constant time and plan in time linear.The simulation results indicate that distributed perception-behavior network can be full integrated into behavior-based system.The environmental context matching method is used to eliminate"perceptual aliasing".Finally,a environmental dynamical exploration method is presented based on"environmental familiarity"to realize the effective exploration of the environment.