Phase II year
1995
(last award dollars: 1997)
The development and implementation of nonlinear, learning control systems for Unmanned Air Vehicles(UAVs) will have three major impacts upon their capabilities,; increased survivability, extended range of operations, and reduced development costs. In the Learning-Optimal Control for Unmanned supermaneuverable Technologies (LOCUST) Phase Ii effort NeuroDyne will extend the successful Phase I results in the utilization of neural network based system identification techniques and optimal control methods to rapidly learn changes in a vehicle's aerodynamic behavior and adapt to sudden meteorological effects while ensuring flight stability. The combined use of neural networks and nonlinear characteristics of the vehicle's dynamic response to increase maneuverability and increase survivability. In this effort, NeuroDyne will leverage current research programs in neural networks, nonlinear control, and unmanned aerial vehicle technology to ensure success ful development of LOCUST. The investigators will further leverage in-house and commercially developed on-board sensors and processing techniques to increase both autonomy and range. LOCUST development will also benefit from the investigator's experience in research and development of neural network based reconfigurable flight control systems for the McDonnell F-15 aircraft and RC UAVs to ensure development of a cost effective, robust adaptive control system.
Keywords: Neural Networks Neural Networks Optimal Conntrol Optimal Conntrol System Identification