Research On Multiple Robots Relative Localization And Formation Control Based On UWB

With the continuous development of automation technology,the multi-robot technology are widely used in service,defense,and rescue fields for better flexibility and robustness compared to single robots.Aiming at the difficulty of obtaining relative poses among robots in the environment without GPS signal and localization base station,and the high cost for multi-robot localization and formation control system configuration,a multiple robots relative localization and formation control method based on UWB is proposed in this thesis.The deployment of low-cost,high-precision UWB sensors improves the localization capability of multi-robot systems in environments with obstructed lines of sight or sparse features,and reduces the configuration cost of multi-robot systems.The integration of relative localization algorithm and formation control method improves the adaptability of multi-robot systems.The specific works of this thesis are divided into the following sections:The first is the UWB-based relative localization method.To suppress UWB Non-Line-ofSight(NLOS)ranging errors,the first channel power of UWB and the total receiving power variation within the sliding window is used to identify the NLOS situation,and the Kalman filter algorithm is used to estimate the UWB ranging results under different environments.And the estimation of relative poses,relative position and heading,between robots are obtained by constructing a nonlinear least squares problem using the odometry-predicted robot poses and the inter-robot range measurements.To suppress cumulative odometry errors within a sliding window,the estimated odometry covariance estimated by extended Kalman filter is applied in a weighted manner to nonlinear optimization.Additionally,a graph optimization algorithm is used to fuse the relative poses obtained by odometry and nonlinear optimization to further improve the localization accuracy and stability.The second is the relative localization-based formation control method.The relative poses between robots are applied to the formation,and the formation is modeled by the leaderfollower method,which generates the virtual target poses of the follower robots through the expecting the formation under the leader coordinate system,and the follower robots track the virtual targets to complete the formation by the linear feedback controller.Besides,the vector field histogram algorithm is implemented for formation dynamic obstacle avoidance in complex environments.When the formation passes through an area with dense obstacles,multiple sets of Li DAR scan data provided by the leader robot and karto SLAM algorithm are used to construct sub-maps,through which the passability of the current situation is evaluated,and the switch between triangular and linear formation shape is completed according to the passability,thus the formation’s adaptation to the environment is improved.The last part is the experiment in a real environment.The results show that the proposed multirobot relative localization method can obtain relative poses with the average position accuracy of 0.32 m and the angular accuracy of 4.16° in a 6m×12m indoor environment.Based on the acquired relative poses and proposed formation control method,steady formation control and flexible obstacle avoidance are achieved during the formation march,under an overall formation tracking error less than 1m.