Pursuit-Evasion Games Using Sensor Networks (MAS 11)

We investigate the differential pursuit-evasion games between multiple pursuers and multiple evaders. Pursuers make use of a hierarchical sensor network embedded in the environment in order to perceive the evader locations. Evaders are localized to some topological nodes using Received Signal Strength Indication (RSSI), which results in a coarse estimate of the evader locations. Based on this information and particle filter based self-localization of the pursuers, we use optimal and sub-optimal pursuit-policies and some task allocation schemes so that the pursuers capture the evaders in a time optimal manner. The effect of sensing errors and other network related errors due to communication delay and packet loss on the pursuit-evasion games are also investigated.