Adaptive UGV Routing via Object Recognition and Analysis (AURORA)

The Army has identified a need to develop a system that can identify, classify, and analyze visual data from unmanned ground vehicle (UGV) and stationary visual surveillance sources to enable real-time decisions and system-wide planning regarding route, speed, and tasks on board UGVs. We propose to design and develop a system for Adaptive UGV Routing via Object Recognition and Analysis (AURORA). AURORA is a distributed software surveillance system that fuses observations from UGV cameras with observations from a fixed (e.g., building-mounted) camera infrastructure to produce a unified view of pedestrians, vehicles, and potential hazards in the operational area. AURORA builds on advanced camera-based pedestrian and vehicle tracking and behavior analysis software designed to operate on both moving and fixed visible and infrared (EO/IR) cameras. The underlying architecture emphasizes distributed processing using Robot Operating System (ROS) nodes to minimize bandwidth requirements. AURORA will build on this architecture by introducing four key capabilities: (1) fusing UGV-based observations with fixed-camera observations; (2) detecting emergent obstacles and hazards, such as potholes, roadwork, or debris; (3) detecting unusual or suspicious behaviors by pedestrian and vehicles; and (4) optimizing and planning routes for a UGV fleet.