A Switching Strategy for Robust Cooperative Exploration

Abstract

This paper develops a switching strategy for networked mobile sensing agents to automatically decide when to perform individual exploration and when to perform cooperative exploration of an unknown scalar field corrupted by time-varying non-Gaussian noises. The switching condition from individual exploration to cooperative exploration is based on Razumikhin theorem, and the change of the signal-to-noise ratio (SNR) serves as the switching condition from cooperative exploration to individual exploration. A cooperative H∞ filter is proposed when the agents are collaborating. Sufficient conditions are derived for the cooperative H∞ filter to converge. This switching strategy balances the mobility of the agents and the exploring accuracy. Simulation results demonstrate that with the switching strategy, a group of sensing agents can succeed in finding a local minimum of an unknown scalar field efficiently.