Improved particle filter algorithm for robot localization

Abstract

For solving the problems of mobile robot SLAM (Simultaneous Localization and Mapping) in unknown environments, this paper presents an optimized RBPF algorithm. The method employs the UKF algorithm instead of the EKF algorithm to estimate landmarks, so it can avoid the derivation of complicated Jacobian Matrix and reduce the error generated by linearizing the nonlinear system. Using the Euclidean distance of particle approximate distribution to the UKF assistant proposal distribution as an adaptive particle-resampling criterion, it can avoid particles' impoverishment and deviation to the real posterior distribution. The experimental results demonstrated these strategies can reduce the localizing complexity and enhance the algorithm's real time speed and reliability.