Multiple agent team theoretic decision-making for searching unknown environments

This paper explores the usage of team theory results to multiple agent search problems. We present a new formulation of a multiple agent search problem that can be solved as a nonlinear optimization problem in a centralized perfect information case and also has features that allows the problem to be reformulated in the framework of a linear-quadratic-Gaussian problem that admits a decentralized team-theoretic solution using Radner's result that equates person-by-person optimality with global optimality. Both the centralized strategy and the team theoretic strategies are derived and some numerical results are presented for illustration. This is the first contribution in the literature that combines fundamental results from search theory and team theory to solve practical problems.