New Monte Carlo algorithm for mobile robot self-localization

A novel Monte Carlo method is proposed aiming at the solution of unmodeled motion problem (such as bumping or kidnapped problem) which is inextricable merely using conventional Monte Carlo localization. By adopting both p(Xkιzk) and p(XkιXk-1) as importance functions and sampling from them, the global localization and kidnapped problems are figured out efficiently. The over-convergence and uniformity validations are introduced to verify correspondence between sample distribution and sensor information for timely resampling which highly saves computational resource and enhances localization efficiency. Experimental results validate the favorable performance of this approach.