Autonomous Exploration for Mobile Robot Based on Graph Topology-Aided Planner
Zhongcheng Peng, Wentao Yu, Kun Yao, Gan Zhang, Xuan Xia · 2025
Autonomous exploration in large-scale, multiple branches unknown environments is a challenging research subject due to low efficiency of map building, large amount of computation and easy deformation of maps. To handle such complex environment, we propose a graph-based topology-aided planner (GBTAP) for mobile robot to realize efficient and accurate autonomous exploration. In order to improve the global planning capability and efficiency of mapping for robots, the graph topology-aided sub-area frontier sampling exploration strategy is proposed. Meanwhile, an active loop closure detection strategy is adopted, which is able to actively visit potential loop closure points to trigger SLAM global pose optimization, thus relieving the map deformation problem. Aiming that the high computational cost of the SLAM global pose optimization module, a hierarchical pose optimization strategy is adopted, transforming the global pose optimization problem into pose optimization between sub-areas and pose optimization within loop closure sub-areas to reduce computational complexity. The experiment results show that the GBTAP algorithm can effectively improve the robot's exploration efficiency and suppress map deformation.