This paper presents a novel method for selecting efficient and robust control strategies required for collaborative operations of unmanned aerial vehicles (UAVs) and unmanned ground vehicles (UGVs) in performing crowd surveillance missions. In this work, a dynamic data driven adaptive multi-scale simulation (DDDAMS) framework is adopted, where a fast running agent-based simulation is used as the strategy maker. Different control strategies are devised during the planning stage for each of 1) team formation, 2) information aggregation, and 3) motion planning of UAVs and UGVs. The devised strategies are then used in the control stage to determine 1) the assignment of different numbers of UAVs and UGVs as a team, 2) the aggregation of low fidelity data from UAVs with high fidelity data from UGVs, and 3) the balance among different objectives (e.g. minimizing the traveling distance, minimizing a change in altitude/elevation) in choosing vehicles' paths. To test, demonstrate, and validate the performance of the method for selecting control strategies, a testbed containing various hardware and software components has been developed. The preliminary results reveal the efficiency and robustness of the proposed approach in terms of the crowd coverage percentage as the system performance.