Inverse estimation of multiple muscle activations under isokinetic condition

Electromyogram (EMG) is a useful technique for recording activation process by measuring a summation of motor unit action potentials (MUAP) produced by muscles. EMG signals have been acquired and processed to establish mappings between muscle activation, torque or/and joint position. A common way for constructing such relationship is to use torque, position and multiple-channel EMG signals. In this paper, all three muscles’ activations are estimated through torque and position information via using multi-output regression. Results show that such approach is effective in multiple muscle(s) activations (EMG) identification and predictions.