The question of how to aggregate autonomous agents with limited abilities in the absence of centralized coordination is known as the Decentralized Gathering Problem. The authors present a bio-inspired aggregation scheme that solves this problem and study a first application of this scheme to a small team of robots. The robots (Alice and Khepera III) obey simple rules and have only a rudimentary perception of their environment. The collective behavior is based on stigmergic principles and uses an active environment to relay the communications between robots. This results in an aggregation process that shows good properties of robustness and that can in principle be extended to swarms of robots.