Research on the Location of Mobile Robot Based on an Improved Monte Carlo Algorithm

Abstract

The particle filter algorithm solves the problem that the stochastic quantity must satisfy the Gaussian distribution for the non-Gaussian filtering. In recent years, it has been widely used in research of target tracking and positioning. The problem of particle scarcity and the choice of the proposal distribution function existed in the robot localization method. In allusion to the two problems, a method of robot localization based on particle filter of Markov Chain Monte Carlo (MCMC) is proposed. By calculating the probability of filtering particles and adaptively adjusting the boundary of the proposal distribution function, it not only maintains the diversity of particles, but also inhibits particle scarcity. The algorithm and the common Sampling Importance Resampling (SIR) algorithm are applied to the simulation experiment of target location. The simulation results showed the good performance of the improved particle filter algorithm, and the phenomenon of article degeneracy can be effectively curbed. Moreover, the positioning accuracy is higher, and the robustness is better than SIR algorithm.