Self-adaptive Pose-tracking Algorithm for Mobile Robots in Dynamic and Highly-occluded Environments

In dynamic and highly-occluded environments such as a cafeteria, metro station etc. the pose-tracking accuracy of mobile robots will be greatly influenced since the occlusion degree and map features are different. To solve this problem, a self-adaptive pose-tracking algorithm is proposed. The main idea is to ensure the accurate and robust robot localization through correcting the odometer-based proposal distribution function (PDF) in particle filter (PF) based on the localizability, which is defined to evaluate the influences of both the dynamic obstacles and prior-map (different structures and uncertainty) on localization. Furthermore, to guarantee the robustness in different dynamic and highly-occluded environments, the reliability parameter between the observations of laser range-finder (LRF) and the measurements of odometer in fusion process is improved. The simulation and experimental results demonstrate that the improved reliability parameter is valid, and the proposed algorithm is accurate and robust for pose-tracking in dynamic and highly-occluded environments.