Development of a Generic 6-DOF Helicopter Controller Model with Variable Time Steps for CHOPPA

The use of Artificial Intelligence (AI) agents to represent helicopter pilots inside closed-loop simulations has been a significant focus of collaborative research activities between DSTO’s Air Operations Division (AOD) and the University of South Australia. CHOPPA (Combined Helicopter OPerations and Performance Analysis) is a MATLAB/Simulink based simulation framework developed by AOD which supports Operations Research (OR) into Australian Defence Force helicopter tactics. This paper presents the research work conducted by AOD and UniSA to develop a novel, generic, 6-DOF helicopter controller model with a variable time-step for use by AI agents within the CHOPPA framework. The 6-DOF helicopter dynamics are modelled by a linearised state-space system with a state vector comprising orthogonal velocity components, roll and pitch attitudes, as well as derivatives of roll, pitch and yaw attitudes. The matrix and vector quantities included in the linearised state-space model are determined for a given time-step in discrete-time simulations. The new state vectors are calculated using the equation of motion. The helicopter mission times of the Monte Carlo runs within the CHOPPA framework can be two to three hours long. Using a fixed time-step for such a long mission time takes up excessive memory and significantly slows down the simulation speed. To improve the computational efficiency of the framework we propose an adaptive time-step algorithm for the 6-DOF helicopter controller whereby the time-step for control vector updates is adjusted based on observed change in the control vector, as well as high-level pilot commands. The new adaptive time-step algorithm produces control vector updates that are less frequent in level flight and more frequent during flight manoeuvres. The effectiveness of the developed controller is demonstrated by simulation studies.