Switched and symmetric pursuit/evasion games using online model predictive control with application to autonomous aircraft

This paper describes a supervisory controller for pursuit and evasion of two fixed-wing autonomous aircraft. Novel contributions of the work include the real-time use of model-predictive control, specifically nonlinear model predictive tracking control, for predictions of the vehicle under control, as well as predictions for the adversarial aircraft. In addition to this inclusion, the evasive controller is a hybrid system, providing switching criteria to change modes to become a pursuer based on the current and future state of the vehicle under control, and that of the adversarial aircraft. Results of the controller for equally matched platforms in actual flight tests against a US Air Force trained F-15 test pilot are given. Extensive simulation analysis of the symmetric games is provided, including regressive analysis based on initial conditions of height advantage, and relative velocity vectors, and in particular the effect of allowing the evading aircraft to switch modes between evader and pursuer during the game.