Distributed Filtering for Time-Delayed Deployment to Multiple Sites

We address the synthesis of distributed control policies for a homogeneous robot ensemble assigned to monitor multiple locations. Our approach uses an appropriate macroscopic description of the ensemble dynamics to first identify spurious behaviors exhibited by the ensemble as a result of robot-robot interactions that arise from operating within a shared environment. This macroscopic analysis is then used to design and improve the agent-level control policies and enhance the overall team performance. In particular, we consider the time-delayed task assignment problem where deterministic (or near deterministic) task execution times result in extraneous robot-robot interactions which degrades the ensemble efficiency. The main contribution is a novel approach towards synthesizing distributed filters to smooth out spurious interactions. We validate our distributed filter both in simulation and in actual robotic experiments.