Active model based predictive control for unmanned helicopter in full flight envelope

For the control of unmanned helicopters in full flight envelope, an active model based control scheme is developed in this paper. An adaptive set-membership filter (ASMF) is used to online estimate both the model error due to flight mode change and its boundary, taking advantage of ASMF, so that the model error can be assumed unknown but bounded (UBB). The proposed approach is practical because the model error depends on both helicopter dynamics and flight states, and may not be assumed as white noise. An active modeling based stationary increment predictive control (AMSIPC) is also proposed based on the estimated model error and its boundary to optimally compensate the model error, as well as the aerodynamics time delay. The proposed method has been implemented on the ServoHeli-20 unmanned helicopter platform and experimentally tested, and the results have demonstrated its effectiveness.