Revisiting the F-8 Aircraft Control Problem with Dynamic Programming

This paper presents empirical research on the implementation of dynamic programming to a classic and established nonlinear dynamical model of an F-8 aircraft’s longitudinal motion. The model has been frequently used as a case study in academic environments for nonlinear dynamics and optimal controls topics. On the other hand, dynamic programming gives an exact solution to an optimal control problem. Hence, it can be used to benchmark other optimization methods. Before implementing dynamic programming, we first investigated the primary behavior of an F-8 aircraft’s dynamical model. The existing dynamical model is inherently stable in the presence of a non-zero initial attack angle as long as its value is less or equal to 31.6°. However, the stall angle of an F-8 aircraft is reported as 23.5°, which is lower than 31.6°. Therefore, in this paper, an initial attack angle that is larger than 31.6° becomes one of our concerns. Our results show that dynamic programming can be used to regulate the states of an F-8 aircraft’s dynamical model even when the applied initial attack angle is larger than 31.6°. We also compare our dynamic programming implementation to the MATLAB pattern search method and sequential quadratic programming. While the first gives results that are very similar to dynamic programming, the latter fails to provide satisfactory results.