Safe learning-enabled and competence-aware autonomy stack
Advanced Air Mobility systems have complex, nonlinear dynamics and are subject to significant dynamic and environmental uncertainties resulting from modeling inaccuracies, physical failures, and external disturbances. They need to perceive the environment, interact with other agents, reason about the uncertainties in perception, and predict other agents’ behavior. Toward this end, our team will focus on the development of a competence-aware perception-planning control framework with learning-enabled controls.
Fast data-driven and code-level verification
We envision that the entire autonomy pipeline will have a mix of modules learned from data and modules programmed by humans. Thus, our goal here is to develop general tools and techniques for rapid and code-level verification of autonomous systems that can meet the challenges of scaling verification to a heterogeneous mix of ML modules with inductive biases and hand-written code.
Runtime fault detection and reachability analysis for in-flight contingency management
All systems are subject to faults; hence, the reliability of Advanced Air Mobility systems and their components mandates the timely detection and identification of faults. Our team targets a comprehensive framework for runtime fault detection and reachability analysis to facilitate in-flight contingency management.
Testing, evaluation, and integration toward technology transfer
The multi-inter-disciplinary nature and the vision toward generating robust, resilient, and certifiable technologies transferable to real-world Advanced Air Mobility and Urban Air Mobility platforms necessitate tight integration between various technologies and tangible efforts with industry partners to prepare the technologies for transitioning into industry. For the transition, we aim to create and leverage existing infrastructure, testbeds, and methods to facilitate the successful integration and fielding of the products of our team’s research. We are uniquely positioned to conduct the required large-scale field testing activities as UNR has the expertise and is authorized to organize, execute, and supervise UAS flight tests up to 1200 ft AGL in the daytime (400 ft AGL at nighttime), within class-G Airspace located at the outskirts of the city of Reno and the broader N. Nevada region.