Robotics for rehabilitation and human amplification is imminent to become part of our daily life. The juxtaposition of human control capability and machine mechanical power offers a promising solution for human assistance and physical enhancement. This work presents an upper limb active exoskeleton controlled by an alternative and simple Human-Machine Interface (HMI) that uses a Hill Muscle Model for strength and endurance amplification. Pneumatic Artificial Muscles (PAM) are used as actuators for its high power-to-weight ratio and to drive the system through a cable arrangement. Genetic Algorithms (GA) approach locally optimizes the model parameters for the actuator mathematical model and the physiologic muscle model that uses the surface electromyography (sEMG) to estimate the exoskeleton joint torque. The proposed methodology offers three main advantages: (i) it reduces the number of electrodes needed to monitor the muscles, (ii) it eliminates the need for user force or pressure sensoring, and (iii) it reduces the real-time processing effort which is necessary for embedded implementation and portability. The exoskeleton is restricted to the right upper limb and the control methodology is validated evaluating the user performance while dynamically and statically handling a 3.1kg payload with and without the aid of the assistive device.