Performance, Precision, and Payloads: Adaptive Nonlinear MPC for Quadrotors

Agile quadrotor flight in challenging environments has the potential to revolutionize shipping, transportation, and search rescue applications. Nonlinear model predictive control (NMPC) recently shown promising results for agile control, but relies on highly accurate models maximum performance. Hence, uncertainties form of unmodeled complex aerodynamic effects, varying payloads parameter mismatch will degrade overall system In this paper, we propose L1-NMPC, a novel hybrid adaptive NMPC learn online immediately compensate them, drastically improving performance over non-adaptive baseline with minimal computational overhead. Our proposed architecture generalizes many different from which evaluate wind, unknown payloads, conditions. The method demonstrates immense flexibility robustness, more than 90% tracking error reduction under large disturbances without any gain tuning. addition, same controller identical gains can accurately fly racing trajectories exhibiting top speeds 70 km/h, offering improvements around 50% relative baseline.