An Adaptive SINDy-Lyapunov Model Predictive Control Framework for Dual-System VTOL UAVs

Abstract

This paper presents an adaptive control framework for dual-system VTOL UAVs capable of operating in both rotary-wing and fixed-wing modes. These aerial vehicles present considerable control challenges due to their nonlinear, time-varying dynamics and inherent instability during flight-mode transitions. The proposed approach addresses these issues by leveraging nonlinear system identification via Adaptive Sparse Identification of Nonlinear Dynamics (ASINDy) with a Lyapunov-based Model Predictive Control (LMPC) scheme. This integrated framework facilitates continuous model updating and guarantees stable trajectory tracking and robust performance. Compared to the GA-PID, the ASINDy–LMPC approach reduced tracking error by approximately 65%, maximum deviation by 67%, average deviation by 79%, and power consumption by 73% in simulation, while nearly halving the control effort. Preliminary hardware trials on a VTOL UAV prototype corroborate these trends, demonstrating consistent improvements during hovering and outdoor flights.