A new MCMC algorithm fusion with distributed perception for mobile robot self-localization

A novel particle filtering localization algorithm which is fused with distributed visual perception based on MCMC method is proposed. For distributed sensor networks composed of several environmental camera nodes and laser sensor on the robot, the entropy-based measurement selection method is adopted which selects the best sensor node to update robot pose, thus highly saves computational resource and enhances localization efficiency and reliability. At the same time, the conventional particle filtering algorithm is fused with Markov Chain Monte Carlo algorithm - Metropolis Hastings sampling which resolved degeneracy phenomenon and the global and kidnapped localization. Experimental results validate the favorable performance of this approach for robot localization.