Behavioral system theory–based direct segmented data-driven predictive control for unmanned aerial vehicles

Behavioral system theory–based direct data-driven predictive control has recently emerged as a promising alternative to model predictive control, eliminating the reliance on explicit physical models. However, its performance is challenged under uncertainties and nonlinear processes. In this paper, we develop a direct data-driven predictive control method based on a segmented prediction trajectory for nonlinear unmanned aerial vehicles, which can improve the control performance in stochastic environments by reconstructing the control horizon trajectory. Experimental results demonstrate that the segmented control method significantly outperforms the unsegmented method under strong disturbances, while both methods achieve comparable performance under weak disturbances. These findings indicate that the selection of data-driven predictive control methods should consider both disturbance conditions and computational efficiency in practical applications.