21st Congress of International Council of the Aeronautical Sciences, Melbourne, Australia, 13-18 September, 1998
Paper ICAS-98-3.7.1

CONTROL OF HIGH ENDURANCE UNMANNED AIR VEHICLE

Wharington J., Herszberg I.
Royal Melbourne Institute of Technology, Australia

Keywords: high endurance, uav

In the quest for very high endurance, designers of Unmanned Air Vehicles may need to develop control systems that extract energy from gust patterns such as thermals. This paper presents control systems for two stochastic, optimal control problems that arise from this concept of soaring UAV: the first involves centering the vehicle about a thermal, and the second involves the use of thermals and inter-thermal gusts. In both cases, reinforcement learning provides the basis for high level, direct adaptive control that drives a low level control system. The low level systems were constructed with the aid of a heuristic and open-loop solution respectively. Simulation results indicate the success of the designs, though faster convergence is necessary for practical application.

view full paper