Abstract:

To solve the problem of the simultaneous localization and mapping (SLAM) under the complex circumstance, a distributed algorithm of the SLAM based on partition of space-region is proposed considering the respective advantages of centralized configuration and distributed structure. The region is formed according to the angle between two landmarks and the robot, which is designed in case of the collinearity between two landmarks. The landmarks in each region are combined to establish the corresponding observation model. Besides, the position of the robot is obtained by applying the distributed unscented particle filter and the positions of the landmarks are estimated simultaneously by employing the Kalman filter. Meanwhile, the accuracy and the stability are improved through constructing the adjustment of particle distribution during the dynamic reconfiguration process. Eventually, the better real-time capability and filter accuracy of the proposed SLAM algorithm are proved through simulation experiments which are supported by actual data.