Robot-assisted therapy has been established as a useful rehabilitation tool for motor recovery in people with various neurological impairments; however, balancing human and robot contribution, such that the target muscle is sufficiently exercised, is necessary to improve the outcome of rehabilitation protocols. Functional Electrical Stimulation (FES) can assist a person to move their limb by contracting the muscle; however, motor assistance is often necessary to accurately follow a desired limb trajectory, especially since stimulation can be limited due to various factors (e.g., subject comfort, stimulation saturation). In this paper, a motor is tasked with intermittently assisting the FES-activated biceps brachii in tracking a desired forearm trajectory whenever the FES input reaches a pre-set comfort threshold. A Lyapunov-like switched systems stability analysis is used to prove exponential stability of the human-robot system. Preliminary experiments demonstrate the feasibility and performance of the controller on two subjects with neurological impairments.