Robotic rehabilitation systems deliver high-dose, high-intensity, and reproducible therapy that augments — not replaces — the licensed clinician. This chapter walks the standards stack a developer or rehab engineer needs when bringing a robotic upper-limb or lower-limb rehab platform to market: where the IEC 80601-2-78:2019 particular-requirements document fits, how the device is classified under FDA 21 CFR 860.3 and EU MDR Annex VIII, and the per-control-strategy options (impedance, admittance, assist-as-needed) you choose between for a given clinical phase.
The reference catalogue spans Lokomat (gait), ArmeoSpring / ArmeoPower (upper limb), InMotion ARM/HAND (Bionik, formerly Interactive Motion Technologies), Bi-Manu-Track (Reha-Stim Medtec), Amadeo (Tyromotion), KineAssist, ZeroG, and pediatric variants. Each platform commits to specific kinematic and dynamic envelopes; the evaluation method follows IEC 80601-2-78 §5 essential-performance requirements with the per-axis position + velocity + force verification battery. The standard does not prescribe a particular control law — it constrains the safety envelope around whatever law the manufacturer adopts.
Impedance control (Hogan 1985) and admittance control are the two dominant paradigms. The choice is driven by patient acuity: stiff impedance early when the patient cannot generate force; soft admittance later when the patient drives the trajectory. Assist-as-needed (Reinkensmeyer + Akoner + Ferris 2002) modulates assistance trial-by-trial against patient performance, preserving the motor-learning signal. Error-augmentation (Patton + Mussa-Ivaldi 2004) deliberately magnifies trajectory error to recruit a stronger correction; it remains a research-grade modality outside specific clinical protocols.
Per IEC 80601-2-78 §201.13, every robotic rehab device implements at least two independent emergency-stop channels and a watchdog-based motion supervisor. The patient-applied-part is type BF or type CF per IEC 60601-1; for FES-integrated robotics, the type-CF requirement applies to the stim channel. Range-of-motion and torque envelopes are configured per-patient against the per-joint anatomical limit and the per-protocol prescription, with hard mechanical end-stops as the ultimate safety boundary.
Pivotal trials follow ISO 14155:2020 + ICH E6(R3) + the per-jurisdiction IDE or equivalent. The primary outcome is typically the Fugl-Meyer Assessment (Fugl-Meyer 1975) for stroke upper-limb, ARAT (Lyle 1981), or the 10-meter walk for gait. The minimal clinically important difference is condition-specific and must be pre-specified in the protocol. Post-market clinical follow-up per EU MDR Annex XIV Part B keeps the evidence current across the device's service life.