Muscle group activity estimation utilizing state observer and neuromusculoskeletal system model

This paper presents a novel inverse muscle activity estimation method utilizing neuromusculoskeletal system model and state observer. The state-space equation of neuromusculoskeletal system is elaborated with standard models in physiology and proposed neural input model. Then, the observability of neuromusculoskeletal system is investigated and is physiologically satisfied. Therefore, the muscle activity can be analytically estimated by a state observer algorithm. An experimental study is conducted to verify the proposed method in the case of normal gait and the estimation result shows reasonable values compare to average EMG patterns. Since models used in the proposed method are purely based on the physiology and only utilize joint kinematics and ground reaction forces to estimate muscle activities, proposed method are expected to provide simple, fast and physiological estimation which will lead to a human-friendly controller design for bio-inspired assist devices.