Multi-agent Pursuit-Evasion Under Uncertainties with Redundant Robot Assignments: Extended Abstract

Abstract

We consider a pursuit-evasion problem with a heterogeneous team of multiple pursuers and multiple evaders. The pursuers (robots), using only noisy on-board sensors, can make a probabilistic estimation of positions of multiple moving evaders based on sensor measurements of signals emitted by the evaders. The evaders being aware of the environment and the position of all pursuers follow a strategy to actively avoid being intercepted. We model the evaders' motion as a time-varying Markov process, and along with stochastic measurements, the pursuers use Markov Localization to update the probability distribution of the evaders. A search-based motion planning strategy is developed that intrinsically takes the probability distribution of the evaders into account. Pursuers are assigned using an assignment algorithm that takes redundancy into account, such that the estimated net time to capture the evaders is minimized. Our experimental evaluation shows that the redundant assignment algorithm performs better than an alternative nearest-neighbor based assignment algorithm.