Functional electrical stimulation (FES) applies low- current high-voltage electrical pulses to muscles to induce a torque-generating contraction. While FES is widely used for movement rehabilitation, it is challenging to control the muscle response for goal-oriented actions, due to the many physical and neural sources of variation in the signal-to-muscle response pathway. This paper aims to describe sources of variation that have not previously been discussed for learning FES control, and proposes how a neuromuscular simulation might leverage this knowledge through domain randomisation to help develop adaptive real-world controllers.     «

Functional electrical stimulation (FES) applies low- current high-voltage electrical pulses to muscles to induce a torque-generating contraction. While FES is widely used for movement rehabilitation, it is challenging to control the muscle response for goal-oriented actions, due to the many physical and neural sources of variation in the signal-to-muscle response pathway. This paper aims to describe sources of variation that have not previously been discussed for learning FES control, an...     »