For unmanned aerial vehicles (UAV) to operate freely in the national airspace system (NAS), they must have an onboard sense-and-avoid (SAA) system capable of providing an "equivalent level of safety" compared to manned aircraft. Among the wide range of candidate sensors for SAA, a passive (i.e., non-emanating) electro-optical (EO) sensor system has weight, size and power advantages for use in a UAV platform. Unfortunately, developing an EO sensor system often requires extensive flight tests for its verification and validation, adding cost and time. Under the Sensing for UAV Awareness (SeFAR) program sponsored by Air Force Research Laboratory (AFRL), Northrop Grumman Corporation developed a real-time EO sensor system simulator to help assess not only the effectiveness of candidate EO detection and tracking algorithms, but also their integration with downstream autonomous avoidance algorithms in a real-time hardware-in-the-loop (HWIL) ground laboratory. In this paper, the authors describe the overall SAA laboratory simulation architecture with an emphasis on the real-time EO system simulator comprising three synthetic scene generators, each simulating a camera, and their corresponding image processing and tracker designs. Simulation evaluation results will be described for the simulated EO sensor system alone and for the overall SAA system performance under a variety of collision scenarios.
展开▼