This paper presents a method for permanent magnet synchronous motor rotor position estimation at low speed with high accuracy. The high-frequency voltage signal is injected into motor stator windings, and the corresponding current signal containing information of rotor position estimation error is demodulated. A Kalman filter is chosen as the rotor position observer to depress the nonlinear effect caused by inverter and improve the estimation accuracy at low speed. Simulation results show that this method is applicable to both steady state and dynamic state estimation of rotor position. With the method proposed, a nsp (TMS320F2812) based position control system is implemented and tested. The experimental results are also presented in this paper, which show that the proposed scheme can achieve high accuracy at low speed.