Omni-vision and Odometer Based Self-localization for Mobile Robot

Abstract

By analyzing the uncertainties in perception models of omni-vision and odometer systems for mobile robot, a landmark-observation-based self-localization method with Kalman filter is proposed, which fuses the data from multiple sensors at successive observation points. Compared with single-sensor methods, it exploits the differences in uncertainty between omni-vision and odometer systems, and consequently improves the self-localization precision of mobile robot. The experimental results show the validity and feasibility of the proposed method.