Probability Based Robot Localization

This paper systematically analyses and studies the localization approaches for autonomous mobile robots navigation and mainly on the probability based localization approaches. The research topics include approaches of robot's pose tracking, Markov localization, Particle Filter and other improved PF method.Two different global localization methods based on Bayesian estimation theory are mainly investigated in the paper. The first one is the SRL(Sensor Resetting Localization)approach which proposed to overcome the robot kidnapped problem. It inherits the advantage of PF(Particle Filter), representing multiple uni-modal distribution instead of calculating probability distribution function explicitly. PF represents the belief by a set of samples,which is drawn from the posterior distribution over the robot's poses. The pose estimation algorithm of PF needs less storage space than position probability grids method, and is with higher accuracy and faster computing efficiency. At the same time sensor resetting operation is added to PF which solves the sample deficient, Sensor Resetting Localization resamples a number of samples based on the sensor data when the most recent sensor reading does not agree with where the robot thinks it is, also means the robot is lost. SRL is robust to modeling errors including unmodelled movements and systematic errors. It can be more quickly recover from kidnap state than traditional PF.ASRL(Adaptive Sensor Resetting Localization)algorithm for mobile robot is proposed based on SRL,It adaptively adjusts the number of new needed samples to avoiding samples deficiency, uses few samples realize real-time robot global localization and double-updates samples by sensor information which guarantees the precision of localization, satisfies the requirement of practical system. In the paper, we particularly discuss the motion model and sensor model for ASRL and compare ASRL to SRL, results from the experiments demonstrating validity and robust of the ASRL algorithm.