Multi-mode Planning for a Low-cost Robot Effective Exploration
Ruijiao Li, Jian Xu, Hongbin Fang · 2023
Exploring unknown environment and creating an accurate map is one of the most important capabilities for a autonomous robot. For a robot to efficiently explore unknown environment, it needs to make optimal decisions to select a proper direction to goals and create waypoints to guide the robot to move towards goals based on gathering information about the environments. This work aims to develop an effective exploration approach for a low-cost mobile robot. We proposed a multi-mode exploration strategy for a robot to work in different scenarios like corridors, rooms, tunnels, open spaces etc. Our approach considers the exploration planning hierarchically-global targets and local targets (also named frontier points) to reduce the cost and improve efficiency. The directions to global targets are restricted by virtual grids and sequenced waypoints of subspace. The local planning is maintained with frontier exploration and sample based planning methods. The planner also obtain the semantic information about environments and obstacles for mapping strategy. Our approach follows the divide-and-conquer thought which makes it possible for autonomous mobile robots to perform exploration effectively with limited computing resource such as tiny modular robots or swarm robots.