Any-Com Multi-robot Path-Planning: Maximizing Collaboration for Variable Bandwidth

We identify a new class of algorithms for multi-robot problems called “Any-Com” and present the first algorithm belonging to that class “Any-Com intermediate solution sharing” (or Any-Com ISS) for multi-robot path planning. AnyCom algorithms find a suboptimal solution quickly and then refine that solution subject to communication constraints. This is analogous to the “Any-Time” framework, in which a suboptimal solution is found quickly, and refined as time permits. The current paper focuses on the task of finding a coordinated set of collision-free paths for all robots in a common area. The computational load of calculating a solution is distributed among all robots, such that the robotic team becomes a distributed computer. Any-Com ISS is probabilistically/resolution complete and a particular robot contributes to the global solution as much as communication reliability permits. Any-Com ISS is “Centralized” in the planning-algorithmic sense that all robots are viewed as pieces of a composite robot; however, there is no dedicated leader and all robots have the same priority. Previous centralized multi-robot navigation algorithms make assumptions about communication topology and bandwidth that are often invalid in the real world. Any-Com allows for collaborative problem solving with graceful performance declines as communication deteriorates. Results are validated experimentally with a team of 5 robots.