The development of a heterogeneous system including aerial and ground robots encompasses the solution of several problems. As a prerequisite, the relative localization of the component of the team must be known. The characteristics of the single components can be exploited in order to improve the performance. While the UAV's have a natural role as supervisors, the ground robots can act as "operative hands" of the system. Hence, the UAV is equipped with a camera in order to obtain relative measurements of the robots on the ground. Conversely, the ground robots has a less rich sensor equipment, limited to only odometry measurements. However, no tagging is implemented since it would pose a major computational burden o the UAV to try to identify multiple different markers, and the ground robot looks all the same from above. Moreover, objects in the environment may look as possible robots. The UAV is therefore not able from the measurements alone to associate an identification number (ID) to the measured robots on the ground. In order to solve the identification problem, we have designed and employed a modification of the standard Probability Hypothesis Density (PHD) filter which, fusing the measurements of the ground robots with their odometries is able to estimate the ID number of the measured robots.