A comparative study of control architectures in UAV/UGV-based surveillance system

The goal of this paper is to study the characteristics of various control architectures (e.g. centralized, hierarchical, distributed, and hybrid) for a team of unmanned aerial vehicles (UAVs) and unmanned ground vehicles (UGVs) in performing collaborative surveillance and crowd control. To this end, an overview of different control architectures is first provided covering their functionalities and interactions. Then, three major functional modules needed for crowd control are discussed under those architectures, including 1) crowd detection using computer vision algorithms, 2) crowd tracking using an enhanced information aggregation strategy, and 3) vehicles motion planning using a graph search algorithm. Depending on the architectures, these modules can be placed in the ground control center or embedded in each vehicle. To test and demonstrate characteristics of various control architectures, a testbed has been developed involving these modules and various hardware and software components, such as 1) assembled UAVs and UGV, 2) a real-time simulator (in Repast Simphony), 3) off-the-shelf ARM architecture computers (ODROID-U2/3), 4) autopilot units with GPS sensors, and 5) multipoint wireless networks using XBee. Experiments successfully demonstrate the pros and cons of the considered control architectures in terms of computational performance in responding to different system conditions (e.g. information sharing).