Cloud-based real-time collaborative mapping and merging method for multi-robot with landmark information

Aiming at the poor performance of a single robot construct the map in large-scale environment, a collaborative mapping and merging method based on landmark information and cloud architecture is proposed. Firstly, a network architecture with local independent control and centralized cloud processing is established. The cloud server processes the data uploaded by each robot in real time. Secondly, a single robot explores the environment to build a map and simultaneously detects landmarks in the environment. Then, the coordinates of landmarks binded to each local map are applied to establish the overdetermined equations to resolve the optimal transformation matrix between maps. Finally, the global map is generated and updated dynamically by the proposed fusion algorithm. The results of experiments in different aspects are showed and analyzed to illustrate the effectiveness and robustness of our method.