In this paper we address the problem of finding an object in a polygonal environment as quickly as possible on average, with a team of mobile robots that can sense the environment. We show that for this problem, a trajectory that minimizes the distance traveled may not minimize the expected value of the time to find the object. We prove the problem to be NP-hard by reduction, therefore, we propose the heuristic of a utility function. We use this utility function to drive a greedy algorithm in a reduced search space that is able to explore several steps ahead without incurring too high a computational cost. We have implemented this algorithm and present simulation results for a multi-robot scheme.