Efficient Autonomous Exploration of Unknown Environment Using Regions Segmentation and VRP.

Autonomous exploration of unknown environments is a critical application scenario in robotics. However, existing studies have strived to generate an efficient tour plan that enables the robots to fully explore the environment. This is due to the greedy strategies and the lack of utilizing the information of the unexplored area. To address the above issues, we propose a new exploration strategy. Our approach involves partitioning the unexplored area into distinct regions and utilizing their estimated workload and spatial structure to construct a region graph. Subsequently, a VRP (Vehicle Routing Problem) planner allocates regions to robots and generates a global tour for each robot in a coordinated manner, which is then refined and provided to the robot’s navigation module. Experimental results demonstrate that that our method outperforms existing approaches in terms of exploration speed, total distance, and overlap ratio.