1 EMG-based Position and Force Control of a Robot Arm: Application to Teleoperation and Orthosis Panagiotis K. Artemiadis Kostas J. Kyriakopoulos Control Systems Lab, School of Mechanical Eng. National Technical University of Athens, 9 Heroon Polytechniou Str, Athens, 157 80, Greece {partem,kkyria}@mail.ntua.gr Abstract— This paper presents a methodology for the control of a robot arm, using electromyographic (EMG) signals. EMG signals from the muscles of the shoulder and elbow joints are used to predict the corresponding joint angles and the force exerted by the user to the environment through his/her forearm. The user’s motion is restricted to a plane. An analysis of various parametric models is carried out in order to define the appropriate form of the model to be used for the EMG-based estimates of the motion and force exerted by the user. A multi-input multi-output (MIMO) black-box state-space model is found to be the most accurate and is used to predict the joint angles and the force exerted during motion, in high frequency. A position tracking system is used to track the shoulder and elbow joint angles in low frequency to avoid drifting phenomena in the joints estimates. The high frequency model estimates, the low-frequency position tracker and a Kalman filter are used to control a torque controlled robot arm in the frequency of 500 Hz. The proposed system is tested both on teleoperation and orthosis scenarios. The experimental results prove the high accuracy of the system within a variety of motion profiles. Index Terms— EMG-based control, teleoperation, orthosis I. I NTRODUCTION As the ubiquity of robots has become evident, the issue of human-robot interface has received increased attention. When the issue of interface comes to the level of control, conventional algorithms are replaced by intelligent systems and decision methods. For this reason, there is an increasing demand for a direct and more natural means of interface between the user and the controlled robot. A possible approach is the use of electromyographic (EMG) signals from skeletal muscles, because of their advantage of being both convenient and natural for the master. Since muscles are responsible not only for moving the human limbs but also for exerting forces to the environment, EMG signals can be proved useful for the control of a robot arm, especially in the cases where the arm is controlled in position and force. Telerobotics is an area that can be mostly benefitted by the use of the electromyogram, by using it as the human- robot interface. Fukuda [1] firstly introduced the idea of teleoperation of a robot arm using EMG signals and a position tracking system. Wrist movement was extracted from EMG signals from the forearm muscles, while a position tracker was used to teleoperate the lower arm of the robot. The authors have also used in the past EMG signals to control a robot arm in planar reaching tasks in [2]. Concerning orthotic devices, a lot of robotic mechanisms intended for either rehabilitation or extension of human ability have been developed during the last decades. As examples of the latter, Kazerooni proposed a new class of robot manipula- tors worn by humans in [3]. Similar orthotic devices for the upper limb have been developed and presented in [4], where EMG signals were used as the main control signal for the exoskeleton providing high intuitiveness. A lot of different methodologies have been proposed for the utilization of EMG signals to control robots. A pattern classifier technique has been used for identifying finger motion based on EMG signals in order to drive a prosthetic hand in [5]. A Hill-based muscle model was used to estimate human joint torque in driving an exoskeleton in [4]. The authors have implemented an algorithm using an auto-regressive moving av- erage with exogenous output (ARMAX) model, for estimating planar motion of the human arm during reaching, in order to teleoperate a robot arm [2]. However, there is limited literature on combined position and force estimation using EMG signals, which is undoubtedly a challenging issue for the control of a robotic device either is remotely teleoperated or coupled with the human arm. In this paper, a methodology for controlling a robot using EMG signals from the muscles of the human upper limb is proposed and tested through real-time experiments. Surface EMG signals are recorded from muscles of the elbow and shoulder joints, while motion is restricted to a plane. A position tracking system is also used for monitoring the human arm motion, in the frequency of 60 Hz. EMG signals, recorded in the frequency of 1 kHz, are used to estimate both human motion and force exerted from the user to the environment in the same frequency. The high frequency motion estimates based on EMG signals are refined in real time using the low frequency measurements of the position tracking system, in order to avoid drifting phenomena, by applying a filtering technique. For the generation of the estimates, an analysis of various models is conducted. The motion and force estimated are used to control the robot arm. The system is tested using two scenarios: initially the arm is remotely operated by the user, while in the second setup, the human arm is firmly cou- pled with the robotic arm at its end-effector. The experimental results show that the efficiency of the EMG-based architecture is adequate for controlling the torque-controlled robot arm with high accuracy, within a variety of situations.