Smooth Transition MMPC for High-Frequency Attitude Control in Quadrotor UAVs

This study offers a new multi-model predictive control (MMPC) framework for quadrotor attitude regulation that performs like nonlinear model predictive control (NMPC) while being computationally efficient like linear model predictive control. NMPC regulates well, but its processing requirements make it unsuitable for high-refresh-rate applications like attitude control. While computationally efficient, LMPC lacks stability and performance in nonlinear dynamics. The proposed MMPC approach uses linear models for quadrotor attitude dynamics and a predictive controller switching mechanism for smooth flight state adaption. To mitigate the 'chattering effect' associated with traditional multi-model methodologies - marked by elevated frequency noise in brushless DC motor performance, considerable peak overshoots, and possible actuator malfunction - our solution implements a smooth-switching mechanism known as adaptive gain scheduling. This modification reduces excessive variations in control input, hence improving stability and prolonging actuator lifespan. Comparative assessments reveal that MMPC attains NMPC-Ievel efficacy while maintaining computational efficiency akin to LMPC, outpacing alternative control methodologies such as incremental nonlinear dynamic inversion, sliding mode control, and conventional LMPC and NMPC.