AUGMENTING THE HUMAN PILOT - AI/ML INFLUENCES ON ONE PILOT TO MANY UNMANNED AERIAL SYSTEM FLIGHT

Current FAA regulations (Part 107.31) prohibit UAS operation by the Remote Pilot in Command (RPIC) beyond visual line of sight (BVLOS). This presents an operational problem as many UAS applications contemplated under Part 107 are reliant on flying long distances, such as infrastructure inspections and package deliveries. Obtaining a 107.31 waiver focuses on considerations for how the RPIC will detect and avoid all other aircraft, people on the ground, obstacles, and structures while ensuring the UAS remains in the operational area and does not exceed the performance capability of the command and control (C2) link. Traditional ground-based surveillance systems are often expensive, difficult to deploy, and have difficulty detecting traffic at the low altitudes common to typical UAS operations. Today's most advanced UAS operational concepts include an RPIC flying BVLOS, one pilot to control many UAS, or a combination of both. To unlock the full potential of UAS, operations will need to shift from remote pilot in the loop control, to automation and eventually autonomy provided by artificial intelligence (AI) and machine learning (ML). Technological and regulatory challenges need to be solved to achieve delegation of decision-making from the human to the machine in a complex safety critical environment of the National Airspace System. This paper describes pathways to achieve this, including an examination of near- and long-term benefits of human/machine teaming for UAS.