The approach that achieves the teleoperation between human muscle signals and the mobile robot is increasingly applied to transfer human muscle stiffness to enhance robotic performance. In this paper, we develop a mobile rollator control system applying a muscle teleoperation interface and a shared control method to enhance the obstacle avoidance in an effective way. In order to control intuitively, haptic feedback is utilized in the teleoperation interface and is integrated with EMG stiffness to provide a large composition force. Then the composition force is implemented with an artificial potential field method to keep the robotic rollator away from the obstacle in advance. This algorithm is superior to the traditional APF algorithm regardless of the required time and trajectory length. The experimental results demonstrate the effectiveness of the proposed muscle teleoperation system.