Using a group of robots in place of a single complex robot to accomplish a complex task has many benefits such as redundancy and robustness, faster completion times, and the ability to be everywhere at once. The applications of such systems are wide and varied: Imagine teams of robots containing forest fires, searching for survivors after a natural disaster, or delivering packages to our doorstep. We’ve been talking about these applications for years, but why aren’t they happening yet?
While current approaches to multi-robot coordination have been successful in structured, well understood environments, they have not been successful in unstructured, uncertain environments, such as disaster response. The reason for this is not due to limitations in robot hardware, which has advanced significantly in the past decade, but in how multi-robot problems are solved. Even with significant advances in the field of multi-robot systems, the same problem-solving paradigm has remained: assumptions are made to simplify the problem, and a solution is optimized for those assumptions and deployed on the entire team. This results in brittle solutions that prove incapable if the original assumptions are invalidated. Unstructured environments necessitate a new multi-robot problem-solving paradigm that relies on diversity of control policies to make multi-robot systems more resilient to changing or uncertain environments. In this talk, I will discuss how my research group is addressing this problem, by working toward this novel paradigm for the creation and design of policies for multi-robot coordination.