Passive Non-Cooperative Intruder State Estimation and Optimal-Feedback Avoidance System for UAVs

In recent years, numerous applications for unmanned aircraft systems (UAS) and autonomous air vehicles have emerged, with even more possibilities on the horizon, such as air taxis and last-mile package delivery. Ensuring safety is crucial for integrating UAS into the national airspace system (NAS), with detect and avoid (DAA) capabilities being a vital safety component. Manned aircraft operations are already incredibly safe, thanks to a century of experience, which is now being applied to UAS. However, a key challenge for unmanned flight is the inability to ``see and avoid'' obstacles. Various DAA systems have been proposed in recent years, each employing different sensor modalities with their own pros and cons. Cooperative systems enable all air vehicles to exchange state information, while devices like Automatic Dependent Surveillance-Broadcast (ADS-B) and Traffic Collision Avoidance System (TCAS) use satellite navigation sensors and transponders, respectively, to broadcast position data. The Airborne Collision Avoidance System (ACAS) led to the creation of the ACAS-XU standard for unmanned aircraft. Despite these advancements, a UAS must also address non-cooperative intruders to ensure a safe flight. In this paper, we introduce the development and integration of a passive collision alert system (PCAS) and vision-based guidance systems, designed to optimally detect and avoid non-cooperative intruders. Hardware-in-the-loop simulation and flight experiments are conducted to demonstrate our system capabilities