In this paper, we consider the problem of cooperatively localizing a formation of networked mobile robots/vehicles in SE(2), and adapting the formation to reduce localization errors. First, we propose necessary and sufficient conditions to establish when a team of robots with heterogeneous sensors can be completely localized. We present experimental measurements of range and bearing with omni-directional cameras to motivate a simple model for noisy sensory information. We propose a measure of quality of team localization, and show how this measure directly depends on a sensing graph. Finally, we show how the formation and the sensing graph can be adapted to improve the measure of performance for team localization and for localization of targets through experiments and simulations.