Power-assisted wheelchairs that assists the driving torque by electric motors are effective for the daily life of elderly people and handicapped people. However, the users with hemiplegia can operate only one side of pushrim and have to adjust the driving direction with their feet. Therefore, this paper proposes a one-handed driving control system of power-assisted wheelchairs for users with hemiplegia. The proposed system estimates the desired position by EMG signal in temporal region when the rider chews with right or left back teeth. The target position is estimated based on the pushing torque and EMG signal by fuzzy inference and the wheelchair is controlled to reach the estimated position. In addition, an operation training system by visual feedback to improve the user's proficiency is also proposed. Some experiment results are provided to show the effectiveness of the proposed driving control system and training system.