Continuous Autonomy Runtime Monitoring to Assure Consistent Operational Safety

Advances in aviation offer significant potential benefits in the transportation of packages and passengers by uncrewed aircraft systems (UAS). These aircraft can improve efficiency, reduce costs, increase safety, and provide new capabilities in various industries. However, before uncrewed aircraft can be widely adopted, they must be shown to be safe. New advances in artificial intelligence and robotics can provide the fundamental components, but they can be challenging to certify using existing standards. Although machine perception and motion planning can be trained to detect and recognize hazards, they cannot be verified to a sufficient level ahead of time, making it difficult to meet aerospace certification standards.nbsp;This Phase I proposal introduces a methodology to incorporate high-performance components for uncrewed aircraft in a way that guarantees a high level of assurance. Known as runtime assurance (RTA), this concept wraps less-trusted functions with high assurance functions. We propose to develop and demonstrate RTA logic as it applies to two capabilities related to critical UAS system failures and proximity to people and property: obstacle avoidance and safe landing, both of which are essential to a safe response during a forced landing event. nbsp;We will demonstrate RTA logic in a modern multi-core processor architecture and validate it through the introduction of seeded faults in the autonomous landing system#39;s software and sensors. This methodology will scale to address all functions necessary for safe flight and provide regulators, insurers, and end users the confidence needed to accelerate adoption.