Fusion Positioning Method of UWB and Odometry Based on Graph Optimization

In recent years, intelligent omnidirectional robots for indoor environment have received wide attention. To improve the accuracy of indoor localization of intelligent omnidirectional robots, this paper proposes a fusion localization method based on graph optimization for UWB(Ultra-wideband) and the odometry of robot. The proposed method first uses the distance between the anchor and the tag measured by UWB to perform least-squares solving to obtain the initial localization value. Then the graph optimization model is constructed based on the UWB range information and odometry displacement information as UWB and odometry constraints, respectively. Then the modified Gaussian Newton method is used to solve all robot positions. The experimental results is carried out, the experiment data show that the localization error is reduced by 30% compared to using only UWB for localization. Compare to the odometry track projection only, the presented method have no significant cumulative error. In addition, the experiment show that the present method has a good robust in positioning predict.