Improvement of the low-speed sensorless position control for an interior permanent magnet synchronous motor (IPMSM), which drives a horizontal axis washing machine drum, extended-flux (EF) model-based PI observer (PIO) system is proposed. This control is designed to estimate the derivatives of the EF components, which are voltage components, from which the rotor position information is extracted. Motor parameters change during the operation because of the load and temperature, and those changes affect the performance of sensorless IPMSM drive, especially at low-speed operation. For satisfactory low-speed operation, observer gains should be carefully selected to reduce the influence of measurement noise and un-modeled dynamics. Linear quadratic (LQR) control method is used to calculate the optimal coefficients of a PIO to improve low-speed performance. The PI observer is chosen for its capability to decouple modeling uncertainties and measurement noise. The effectiveness of the proposed sensorless method is verified by the simulation on MATLAB/Simulink and experimental results in a 900W IPMSM drive for washing machine. The proposed estimator achieves stable operation at 87 rpm 100% and 45 rpm 25% load torque conditions. It has a much better dynamic performance and improved capability and is robust for speed and torque variations in the system.