Abstract: In order to solve the problems of the conflict of high precision and fast response, and stability of robotic arm servo control in a dynamic environment, several measures were proposed. Firstly, taking the permanent magnet synchronous motor (PMSM) as the study object, a radial basis function (RBF) field-oriented control (FOC) system was developed to improve the controller structure, to overcome the integral hysteresis of the PI controller and to improve the response speed of the system. And then, a supervised learning method was used to solve the instability problem of neural network in the control system. An online learning method was applied to improve the adaptability of the control system in a dynamic environment. The experiment results show that the proposed methods can effectively improve the stability of the RBF-FOC system, the dynamic response speed and anti-interference ability of the PMSM.