This paper investigates a new approach to the rapid control of an upper limb exoskeleton actuator. We used a mathematical model and motion measurements of a human arm to estimate joint torque as a means to control the exoskeleton’s actuator. The proposed arm model is based on a two-pendulum configuration and is used to obtain instantaneous joint torques which are then passed into control law to regulate the actuator torque. Nine subjects volunteered to take part in the experimental protocol, in which inertial measurement units (IMUs) and a digital goniometer were used to measure and estimate the torque profiles. To validate the control law, a Simscape model was developed to simulate the arm model and control law in which measurement data from IMUs and a goniometer were fed into the suggested Simscape model. The arm torque profiles are key to the control approach and should be traced by torques produced by the exoskeleton actuators to provide comfort and flexibility for the subjects. A DC motor was used as an actuator for the exoskeleton, and its model was used in the physical Simscape model. To reduce the error in the driving torque compared with the reference arm torque, a PID controller was implemented. The results show the potential of our methodology for tracking and controlling the actuator’s torque, in which the mean square error was reduced to less than 0.2 - a significantly low value.