Gators: a Game-Theoretic Tool for Optimal Robot Selection and Design in Surface Coverage Applications

Over the past several decades, the number of commercially available robotic systems has increased significantly to meet the rising demand driven by theoretical advances and emerging practical applications. Although this growing variety offers more choices to users, it can also be overwhelming as they navigate many options to find the best system for their needs. This market saturation also forces robot providers to ensure that new robots are competitive with or superior to existing systems to increase the economic viability of their products. This need is further complicated in multi-robot applications, where understanding individual contributions to overall team performance is complex but necessary. To assist in task-driven selection and design of capable robotic systems, this paper introduces GaTORS, a novel tool that frames the robot task allocation process as a collaborative, general-sum, discrete-time game. By parameterizing robots with a set of common constraints, GaTORS enables performance evaluation of existing and hypothetical robotic systems to select teams of systems most capable of achieving a given task. We focus on robotic surface coverage applications and apply GaTORS to the problem of surface coverage for corrosion mitigation in an industrial refinery, where it is used to select a team of robots best suited to repair all identified material. We also demonstrate how GaTORS can be used to set targets for system design to create new robots that can outperform alternative systems in assigned tasks. Due to its flexibility, GaTORS can be adapted to provide similar insights for other types of robots in new environments and surface coverage applications. We release GaTORS' code ¹https://github.com/UTNuclearRoboticsPublic/gators open-source.