Hierarchical flight control system synthesis for rotorcraft-based unmanned aerial vehicles

Dr. David Shim, University of California, Berkeley, Department of Electrical Engineering and Computer Science

This talk presents the latest research results of the hierarchical guidance and control system design for rotorcraft-based unmanned aerial vehicles (RUAVs) at University of California, Berkeley. As a part of the comprehensive research developing a group of intelligent autonomous agents, we aim to provide a systematic approach to construct RUAVs with autonomous flight capability using commercially available radio controlled small-size helicopters. Among many aspects of RUAV construction, the issues of autopilot design are introduced. The vehicle dynamic model is identified by applying a time-domain identification to the flight data. With the identified vehicle model, two different control theories are applied: classical SISO control and modern MIMO linear robust control theories. In the classical control approach, the original MIMO system model is decomposed into four SISO subsystems, for which individual classical controllers are designed. Alternatively, a MIMO attitude controller is designed using mu-synthesis control theory. All of these proposed controllers are tested in actual flight tests. Once a satisfactory controller is found and validated during flight tests, low-level controller is integrated with middle-level supervisory layer. This layer is responsible for the guidance of the host vehicle through the given waypoints. The middle layer, often called the waypoint navigator, achieves this goal by switching adequate controllers and generating the reference trajectories. In addressing the waypoint navigation problem, a novel framework called Vehicle Control Language is developed as a mission-independent script language to specify the given flight pattern. The autonomous RUAV developed so far is then applied to "pursuit-evasion game", which is a showcase of the intelligence, vehicle guidance & control, and high QoS realtime wireless network.