Multi-Sensory Fusion for Mobile Robot Self-Localization

In this paper, a novel computation method named MMK (Multi-sensory Markov-Kalman) localization, is proposed for self-localization of mobile robots in polygonal environments. The method combines multimodal robust Markov localization and unimodal efficient KF method, and is much improved by utilizing multi-sensory information. The localization process requires less memory and lower resolution discretization of the pose space. Both efficiency and precision can be improved by using information from sonar and visual sensors. Experiments are conducted to demonstrate the validity of the proposed approach