In order to improve the autonomous flight performance of unmanned helicopters under various operating conditions, this paper presents a nonlinear model predictive controller deployed on an unmanned airborne embedded platform. Because of the computational power requirements of NMPC controllers, the current utilization on UAVs is mainly deployed at ground stations and control the UAVs through communication links, making the UAVs lose part of their autonomy. The objective of this paper is to improve and streamline the conventional NMPC and implement it on embedded platforms. Using an inner and outer loop cascade control architecture, the NMPC is used as an outer loop controller and combined with the NDOB to form a composite control approach. The horizon is increased without increasing the computational effort by using multiple shots at unequal intervals. The experiment successfully deploys NMPC on an embedded platform on an unmanned helicopter and conducts flight tests, demonstrating significant performance advantages.