This work presents a control system for exoskeletons that utilizes electrical signals from the muscles as the main means of information between the human operator and the exoskeleton. Those signals are picked up from the skin on top of selected muscles and reflect the activation of the observed muscle. They are evaluated by an algorithm to drive the control system of the exoskeleton. A calibration algorithm for those parameters is presented which relies on the same algorithm which measures the muscle parameters every time before the machine is used. The calibration is done to eradicate many erratic signal variations.The model fuses the results from different biomechanical and biomedical research groups and performs a sensible simplification considering the intended application. An exoskeleton for the upper extremity was designed and constructed to verify the model and investigate the interaction between the human operator and the machine in experiments with force support during everyday movements.