This paper describes the implementation of a decentralized architecture for autonomous teams of aerial and ground vehicles engaged in active perception. We provide a theoretical framework based on an established approach to the underlying sensor fusion problem [3]. This provides transparent integration of information from heterogeneous sources. The approach is extended to include an information-theoretic utility measure that captures the task objective and robot inter-dependencies. A distributed solution mechanism is employed to determine information maximizing trajectories and assignments subject to the constraints of individual vehicle and sensor sub-systems. This architecture enables the benefit of the complementary aerial and ground based vehicle and sensor capabilities to be realized. The approach is applied to missions involving searching for and tracking multiple ground targets. Experimental results for vehicles equipped with cameras are presented. These illustrate the impact of the team configuration on overall system performance.