Observation-based Cooperation in Mobile Sensor Networks: A Bio-Inspired Approach for Fault Tolerant Coverage

We consider the problem of dispersing nodes of a mobile sensor network to cover an unknown environment. Approaches that use observation to infer state and intent to disperse robot nodes are presented. Nodes use their observations to decide on their next actions. With simulated and physical node experiments, we compare observation-based approaches to no communications, direct communications, and potential field approaches. Experimental results show that using observation to infer state and intent to disperse nodes performs better than non-communicative and potential field based cooperation and in some cases direct communications.