Estimating Angular Joint Positions Based on Electromyographic (EMG) Activity

This study aims to estimate the movement of the upper limb using signals collected by surface electromyography (sEMG). Signals recorded for selected shoulder and elbow muscles during limb motion in the sagittal plane were used. An artificial neural network with fuzzy logic was applied to process sEMG signals. The network forecasts joint motion trajectories. These data are then used to determine joint torques based on inverse dynamics model. The angular trajectories obtained in dynamic simulation are compared with the real ones, allowing for the assessment of the motion estimation accuracy.