Improving The Robot Localization Accuracy Using Range-only Data: An Optimization Approach

This paper proposes a methodology to improve the robot localization accuracy with only range measurement data for wireless sensors, e.g. Wi-Fi, Bluetooth, mobile network base station, ultra-wide band (UWB) tag, or even cameras and lidar with landmark range data. Accordingly, we modeled the robot localization as an optimization problem by using only the range data. Intuitively, this is a nonconvex problem which arises the difficulty of getting the optimal solution. Therefore we further reformulate it as a convex problem and the global optimal solution is guaranteed. Furthermore, the comparison experiments of a robot implementing a traditional Kalman filter using smoothing data as well as our proposed method have been done with the motion capture system benchmark. Both the simulation and field experiments validate the robustness and accuracy of our proposed method. The quantitative evaluation of the localization accuracy increases by 61.29% in simulation and 39.03% in field experiment to the best within a workspace of 4m ×4m in the scenarios.