Reliable and robust autonomy under uncertain environments will be critically important in expanding the capabilities of aerial vehicles in general and autonomous rotorcraft in particular. The proposed activities under this project are based on the premise that autonomy architecture for aerial vehicles will be built through the integration of artificial intelligence into traditional (low level) guidance, navigation and tracking control (GNC) mechanisms. Towards this goal, this project has the following objectives: (1) collision avoidance (sensing and navigation) strategies for moving and emergent obstacles, (2) algorithms for robust operation of autonomous aerial vehicles under sensor degradation, and (3) contract-based design paradigms for artificial intelligence-in-the-loop systems for autonomous aerial vehicles. In order to address these objectives, this project has three sub-tasks (A1 – A3). Subtask A1 focuses on the design of reactive control algorithms for collision avoidance with moving or emergent obstacles in the environment. Subtask A2 addresses robust operation of AAV’s when there is significant degradation of sensory perception. Subtask A3 will create a framework based on assume-guarantee reasoning, in which the performance of the feedback control system is assured through the establishment of contracts for the control components. We aim to demonstrate the results in indoor experiments with micro UAVs as well as through simulations of larger autonomous aerial vehicles (AAV’s) in the 200-2000lb weight range operating in outdoor environments.
