The Application of Multi-sensor Information Fusion by Improved Trust Degree on SLAM

This paper presents a novel multi-sensor information fusion method by improved trust degree for Simultaneous Localization and Mapping (SLAM) on segment-Based maps. The nearest neighbor method is utilized to detect homologous features, and a fuzzy-index belief function is defined to obtain correlation of features detected by various sensors. Then, an objective weight of sensor data Based on trust degree is designed. The paper combines an objective weight with an expert weight to be a fusion weight for acquiring accurate environment features. Finally, Extended Kalman Filter (EKF) is adopted to update robot pose and map. The experimental results show that the algorithm can highly improve the precision of the robot pose and the map.