Lateral Guidance Control of UAV Using Feedback Error Learning

[Abstract] This paper presents an on-line navigation and adaptive flight control for waypoint and line tracking of Unmanned Aerial Vehicle (UAV) using neural network (NN). This system consists of two parts, a navigation system for making rolling angle command from the relative position of the aircraft and the target, and the Feedback Error Learning (FEL) part for rolling angle control. For a navigation system of waypoint tracking, Proportional Navigation (PN), a guidance law which dictates the aircraft to rotate at a rate proportional to the rotation rate of the line of sight, is used. In line tracking, the rolling angle command is created by the deviation from the target line. FEL is a learning scheme of NN which is added parallel to a conventional feedback (CFB) controller. By minimizing CFB signals through a learning process, the NN obtains approximate inverse dynamics of a desired plant. It thus contributes to the improvement of the control performance including the time delay in response and overshoot. Moreover, the adaptability of NN enables the entire control system to show good performance against disturbance such as wind gust and change in system dynamics such as actuator failure. This paper applies a linear NN with a single layer for FEL, because the control system must have high convergence rate and must be simple for a small UAV. A simulation is performed using this method. The results show that the proposed navigation control system is able to improve the whole control performance.