An Efficient Multi-Robot 3D SLAM Algorithm

In this paper, an efficient multi-robot 3D SLAM algorithm is presented for environment reconstruction. A team of mobile robots, each equipped with a Kinect sensor, cooperate to explore the unknown environment with unknown poses. Each of them is allowed to build its local map and to share the local map among other mates. A survival of the fittest strategy that the points and keyframes of the reconstruction are selected to improve the robustness. By matching the overlapping tendency of any two mates, a map fusion strategy is introduced which in result is building a global map. Finally, G2O optimization library is applied to build an optimized global map. Experiments are performed to demonstrate the efficiency of the proposed approach.