Evolving and Certifiable Autopilot for Unmanned Aerial Systems

An intelligent flight control system is developed with learning capabilities and a high degree of assurance that can be certified by the FAA and tested on a modular reconfigurable UAS. Existing lack of intelligence, adaptability and high performance of current automatic flight controllers is addressed by taking advantage of high-performance computing platforms, state-of-the-art machine learning and verification algorithms to develop a new intelligent, adaptable and certifiable flight control system with learning capabilities. The autopilot system will be able to learn from flight experience and develop intuition to adapt to a high level of uncertainties. To provide a high degree of assurance and make the learning autopilot system safe and certifiable, a conventional autopilot system is integrated based on a run-time assurance architecture. A monitor is developed to check aircraft states and envelope protection limits and handover aircraft control to a conventional autopilot system if needed. Provable guarantees of the monitor and the controllers is provided using formal analysis. The hybrid flight control system has adaptability and intelligence of skilled pilots and is capable of performing complex analysis and decision making in real-time. An artificial neural network model is built and trained to mimic the performance of classical robust optimal controllers, extending robustness, adaptability and curiosity of artificial neural network controllers and integrating a Real-Time Assurance system. Technology demonstration of the intelligent flight control system is achieved by flight testing of a Modular Air Vehicle, where the configuration can be customized to fit flight test needs and test adaptability of the proposed technology. A Modular Air Vehicle is designed and prototyped. Once the intelligent flight controllers are integrated with the airframe, ground and flight tests will be carried out to verify the performance and reliability of the proposed technology. More »