Exoskeleton actuators face a unique constraint that humanoid robot actuators do not: they must move in parallel with a human body. This creates the concept of "mechanical transparency" —the ability of the actuator to feel invisible to the wearer. This guide covers every technical requirement from this fundamental constraint outward.
1. Mechanical Transparency: The Defining Requirement
Mechanical transparency describes how naturally an exoskeleton moves with the wearer —an actuator is "transparent" when the wearer feels no resistance from the mechanism itself, only the assistive or resistive force being intentionally applied.
Wearer fights the mechanism. High reflected inertia and friction. Common with high gear-ratio actuators without proper force control.
Some resistance felt. Acceptable for rehabilitation where partial assistance is the goal. High-ratio planetary with torque sensing.
Actuator feels invisible. Required for augmentation exoskeletons. QDD (quasi-direct drive) with low gear ratio < 20:1.
The two primary metrics for quantifying transparency:
- Reflected inertia: Ireflected = Imotor × N² —lower gear ratio N dramatically reduces perceived inertia
- Coulomb friction at output: τfriction = τmotor_friction × N —same issue; lower N reduces output-side friction
2. Quasi-Direct Drive (QDD) Topology
QDD actuators use very low gear ratios (typically 6:1-15:1) with high-torque density motors to achieve the right balance between transparency and torque output. This differs from the high-ratio approach (50:1-160:1) used in cobots.
| Topology | Gear Ratio | Transparency | Torque Density | Best For |
|---|---|---|---|---|
| Direct Drive | 1:1 | Highest | Very Low | Research, ultra-light applications |
| QDD (Quasi-Direct) | 6:1 - 15:1 | High | Good | Exoskeletons, humanoid legs |
| SEA (Series Elastic) | High ratio + spring | Good (measured) | Medium | Rehab exoskeletons, slow tasks |
| High-Ratio Geared | 50:1 - 160:1 | Low | High | Cobots, precision arms |
ZHR-P series with 7.75:1 to 10:1 ratios sits squarely in the QDD range, providing the transparency required for exoskeleton applications while maintaining the torque density needed to support human loads.
3. Joint Torque Requirements: Lower-Limb Exoskeleton
Based on biomechanics data for a 75 kg user during level walking and stair climbing (the two design-driving scenarios):
| Joint | Level Walking (Nm) | Stair Climbing (Nm) | Design Target (Nm) | ZHR-P Model |
|---|---|---|---|---|
| Hip (pitch) | 40-60 | 70-90 | 80 Nm | ZHR-P120 (120 Nm) |
| Knee | 60-80 | 100-130 | 120 Nm | ZHR-P120 (120 Nm) |
| Ankle | 80-120 | 100-150 | 100 Nm | ZHR-P120 (120 Nm) |
Note: Ankle torques appear high because of the ground reaction force moment arm. Design target includes 1.25-1.5× safety factor over peak walking torque.
4. Upper-Limb Exoskeleton Requirements
Upper-limb exoskeletons (shoulder rehabilitation, industrial lift-assist) have lower torque requirements but stricter transparency and accuracy demands —the arm must precisely follow the wearer's motion intent.
| Joint | Required Torque | Backlash Tolerance | Recommended |
|---|---|---|---|
| Shoulder flex/ext | 20-40 Nm | <5 arcmin | ZHR-H17 (harmonic) |
| Elbow | 10-30 Nm | <3 arcmin | ZHR-H14 (harmonic) |
| Wrist | 3-5 Nm | <1 arcmin | CyberGear (QDD) |
5. Control Bandwidth & Safety
Exoskeleton control loops must handle unexpected human motions (stumbles, sudden stops) faster than human reaction time (~200 ms). The actuator and control architecture must provide:
Looking for actuators that actually meet these Nm/kg benchmarks?
Check out the ZHR-H Series (up to 122 Nm/kg) with <5 arcsec backlash. Available for OEM sampling.
Looking for actuators that actually meet these Nm/kg benchmarks?
Check out the ZHR-H Series (up to 122 Nm/kg) with <5 arcsec backlash. Available for OEM sampling.
6. ZHR-P Selection for Exoskeleton Applications
Designing an Exoskeleton—Get Joint-Level Specifications
Share your user body weight, assisted joints, and motion scenarios. Our team will recommend the optimal QDD actuator combination with transparency analysis.
View ZHR-P Specifications