A Control Scheme to Minimize Muscle Energy for Power Assistant Robotic Systems Under Unknown External Perturbation

This paper proposes a novel control method to minimize muscle energy for power-assistant robotic systems that support the intended motions of a user under unknown external perturbations, using surface electromyogram (sEMG) signals. Conventional control methods based on force/torque (F/T) sensors have limitations to detect human intentions and could, presumably, misunderstand or distort such intentions because of external perturbations of the interaction forces, such as those found in activities of daily living. F/T sensors measure the sum of the applied force, including unknown external forces and human intention; thus, a power-assistant robot controller cannot exactly decompose the real human force. In this paper, we describe a counterexample that cannot be supported by conventional force-sensor-based control methods. We also verify why these control methods may guide human behavior in the wrong direction, and thus, have limitations under unknown external perturbations. We then propose a new control method to minimize the muscle energy indicated by sEMG signals. The proposed control approach is fundamentally based on the concept of power-assistance, in which a robot can reduce the users expended muscle energy while performing given tasks. The proposed control approach is verified through experiments using a power-assistant robotic system for the upper limbs under external perturbations.