Neurotech
Prosthetics, exoskeletons and neural control
Controlling an artificial limb from nerve or muscle signals is a solved problem in the laboratory and a much harder one at the clinic.
Neural interfaces reach patients most often not through brain implants but through the peripheral nervous system, in prosthetic control.
How myoelectric control works
The mainstream approach reads electrical activity from residual muscles at the skin surface.
Contracting a muscle produces a signal that electrodes detect. Two electrode sites can drive two functions — opening and closing a hand, or switching between grips.
Conventional devices have used exactly that: a small number of sites, a small number of commands, with mode-switching to access more functions.
It works, it is robust, and it is slow and unintuitive. Users describe switching modes to change grip as the main frustration.
Pattern recognition
The step that improved things substantially.
Instead of mapping one electrode to one function, an array of electrodes captures a pattern of activity across the residual limb, and a classifier learns to associate patterns with intended movements.
This allows more functions without mode-switching, and control that feels more natural because the user thinks about the movement rather than about operating a control.
It requires training data from the user, and the classifier degrades when electrode positions shift, when the limb changes volume through the day, or when arm posture changes — because the muscle activity pattern for the same intent differs by posture.
Recalibration is therefore a routine part of use, and reducing its frequency is an active research goal.
Targeted muscle reinnervation
A genuinely elegant surgical technique.
Nerves that once controlled the missing limb are surgically redirected to remaining muscles — commonly in the chest or upper arm.
Those muscles then contract when the patient thinks about moving the missing hand, providing intuitive control signals that map directly to intended movements.
A related benefit is that reinnervation frequently reduces phantom limb pain, which is a substantial clinical outcome independent of prosthetic control.
A further development redirects sensory nerves to a patch of skin, so touching that patch is felt as touch on the missing hand — a route to sensory feedback.
Sensory feedback
The most important missing capability.
Without feedback, a user must watch the hand constantly to know how hard it is gripping. Grip force is managed visually, which is exhausting and unreliable.
Approaches include vibrotactile feedback on the residual limb, direct nerve stimulation through implanted cuff or intrafascicular electrodes, and the reinnervation route above.
Studies with implanted peripheral nerve stimulation have reported restored touch sensation, improved fine control, reduced phantom pain and — notably — increased embodiment, the sense that the prosthesis is part of the body.
That embodiment finding matters because it correlates with whether a device is used at all.
Abandonment
The number the field takes most seriously.
A substantial proportion of upper-limb prostheses are abandoned by their users. Reported rates vary widely with device type and population, and the figure is consistently high enough to be the central problem.
The reasons are consistent: weight, discomfort at the socket, lack of feedback, unreliable control, maintenance burden, and the fact that many tasks are performed faster one-handed than with a device that requires attention.
Which means a more capable hand is not automatically a better outcome. Reducing weight and improving socket comfort frequently deliver more benefit than adding degrees of freedom.
This is a useful corrective to a field that measures progress in actuated joints.
Lower limb and exoskeletons
A different problem, because walking is more stereotyped than manipulation.
Powered lower-limb prostheses use terrain detection and gait-phase estimation to adjust behaviour for stairs, slopes and level ground.
Exoskeletons split into two categories.
Rehabilitation and mobility devices for people with spinal cord injury, which have clinical approval and are used mainly in supervised settings. Walking speeds remain slow and crutches are generally required.
Industrial exoskeletons, largely passive, supporting the back or arms during repetitive lifting or overhead work.
These are deployed commercially and the evidence on injury reduction is developing rather than settled — they reduce load on some structures and may transfer it to others.
The honest summary
Control from the peripheral nervous system works well enough that control is no longer the main limitation.
Weight, comfort, feedback, reliability and cost are, and they receive far less attention than they warrant relative to their effect on whether the device is worn.
