Comparative Analysis of Single and Multi-Robot Goal Selection for Autonomous Exploration and Mapping

Lu, Li-yi, Kevin Barry, Mohan Krishnan, and Mark Paulik

Robots are increasingly being used for exploration purposes. This includes exploring Mars and the Moon, as well as the bottoms of the oceans. The exploration process requires two components, sequential goal selection and mapping. Mapping algorithms record coordinate references and shape data for observed information, and goal selection algorithms guide the robot to unknown areas.

Exploration can be done by a single robot or by a team of robots. Working together the robots can cover more ground in less time if properly coordinated. A common exploration algorithm known as “Frontier” has many variants, which include market-based goal distribution, revenue-based goal selection, and value-iterative cost-based goal selection. This work presents a comparison of these methods when used in isolation or in combination for optimal exploration in an unstructured outdoor environment.