Based On Improved Particle Filter For Mobile Robot Localization

Position of autonomous mobile robots is the basic problem, particle filter and the mobile robot motion and perception models combine to form the Monte Carlo Localization method can effectively solve the mobile robot in unknown environment positioning. This paper proposed an improved particle filter based on the basic particle filter algorithm.And applied the improved particle filter to mobile robot localization, compared to the basic particle filter algorithm, the improved algorithm in the following improvements have been made:1.Combine the particle filter with a typical MCMC (Markov Chain Monte Carlo) method-Metropolis Hastings (MH) sampling algorithm, improve the basic particle filter algorithm, strive to ensure the particles effectiently and increase the diversity of particles, While improving the accuracy of the algorithm.2. In the sampling phase, we use the EKF (extended Kalman filter) and UKF (Unscented Kalman Filter) algorithm to generate the importance density function, So that the actual distribution and sampling points are close to the same, solve the particle degradation better. While combine the improved algorithm with MH sampling algorithm and simulation analysis, experimental results show that the improved algorithm has better estimation accuracy.3. The improved algorithm is applied to simulation experiment of mobile robot global localization. Simulation results show that the improved algorithm can achieve a better mobile robot global localization.