The maturation of autonomy for electric vertical take-off and landing aircraft will soon make it possible to execute military intelligence, surveillance, reconnaissance (ISR) missions aboard crewed autonomous aerial vehicles. This research experimentally investigates factors that may influence the quality of interaction (i.e., team fluency) between a non-pilot human operator and the AI pilot responsible for autonomous flight, aboard a minimally crewed aircraft. In a flight simulator study with twenty-seven participants, various levels of workload and AI pilot capabilities are investigated including run-time assurance through control barrier functions (CBFs). CBFs are used to enable proactive collision avoidance behaviors by the AI pilot. Team fluency and mission effectiveness outcomes through trust, situation awareness, workload, interaction, and performance show that task complexity and AI behavior are significant factors for the quality of human-AI interaction in the autonomous ISR context.
Link to the full paper: Run Time Assurance and Human AI Fluency in Crewed Autonomous Intelligence Surveillance and Reconnaissance | AIAA Aviation Forum and ASCEND co-located Conference Proceedings