Small size unmanned helicopter has found diverse applications for both civil and military missions. In order to develop a flight robot based on a small size unmanned remotely controlled helicopter, a flight control system is presented in this paper. The hardware and software of the system are designed. The design goal is to have the helicopter achieve stable hover. As the kernel of the system, the flight control strategies are described which based on artificial neural network and human operator's skill. The processed flight data are used to train and validate a multilayer artificial neural network offline, and then the trained artificial neural network is reconstructed in the flight control system for control strategies.