Exoskeleton Torque Explained: Why a Higher Nm Rating Does Not Automatically Mean Better Walking Assistance

When you compare walking-assistance exoskeletons, the first specification most datasheets highlight is torque in Nm. A device rated at 24 Nm sounds more powerful than one rated at 12 Nm — and in a narrow engineering sense, it is. But peak torque alone tells you almost nothing about what a wearer actually feels during a walking cycle, how much assistance arrives at the right moment in the gait phase, or whether the device fits the intended user and use case. A 12 Nm consumer hip exoskeleton with well-tuned gait-phase timing can reduce metabolic cost more effectively than a higher-torque device that delivers assistance at the wrong time. This article breaks down what the Nm number actually measures, which variables matter more than peak torque, and what buyers should verify before comparing devices.

What the Nm Number Actually Measures

Torque, expressed in Newton-meters (Nm), is a measure of rotational force. In an exoskeleton, the Nm rating typically refers to the maximum torque the actuator can produce at a specific joint — most commonly the hip, knee, or ankle. This number is generated by the motor and transmission system: motor torque multiplied by the gear ratio, minus efficiency losses through the reducer.

But this raw mechanical output is not the same as the torque the wearer’s joint experiences as assistance. Between the actuator and the human body, several factors reduce or modify the delivered torque: the compliance of the cuff and strap interface, the alignment between the robot joint and the biological joint, soft tissue displacement, and the control system’s decision about how much torque to command at any given instant. The delivered torque at the joint under real loading conditions is always lower than the actuator’s peak rating, and only arrives during a fraction of the gait cycle.

That distinction is why two exoskeletons with the same peak Nm rating can produce noticeably different walking experiences. The rating describes a ceiling, not a guarantee.

Peak Torque vs Continuous Torque vs Delivered Assistance

Three different torque values appear in exoskeleton specifications, and confusing them leads to mismatched expectations.

Peak torque is the maximum instantaneous torque the actuator can produce. It is sustainable only briefly before thermal protection or current limiting engages. A 24 Nm peak rating means the motor-gearbox combination can briefly reach that output — not that it delivers 24 Nm continuously throughout a walking session.

Continuous torque is the output the actuator is rated to sustain under specified thermal, duty-cycle, and environmental conditions without overheating. This value is always lower than peak torque, often significantly so. For quasi-direct drive actuators used in portable hip exoskeletons, the ratio of peak to continuous torque depends on the motor’s thermal characteristics and the gear ratio selected by the designer. A device designed for brief, high-torque bursts (e.g., sit-to-stand transitions) may have a high peak-to-continuous ratio, while a device designed for steady-state walking assistance may be tuned for a narrower gap.

Delivered assistance is the torque the control system actually commands and the wearer’s body receives during a specific phase of gait. This is determined by the control strategy — not the hardware ceiling. A well-tuned controller may deliver a fraction of the peak rating at the hip flexion phase and 0 Nm during swing, even on a device with a much higher peak rating. The control system decides when, how much, and in which direction to apply torque based on gait phase detection, intention estimation, or pre-programmed profiles.

Why Timing in the Gait Cycle Matters

The human gait cycle has distinct phases — stance, swing, push-off, heel strike — and each phase places different demands on the hip, knee, and ankle joints. Effective exoskeleton assistance depends on delivering torque at the right phase, not just delivering more of it.

Research on task-agnostic exoskeleton controllers illustrates this clearly. A Nature 2024 study demonstrated that a controller estimating the wearer’s biological joint moment in real time achieved 5.3–19.7% metabolic cost reduction across 10 different activities — walking on flat ground, inclines, declines, and other locomotion modes — without requiring activity-specific tuning (Nature, 2024; S-tier source). The key was not high peak torque but adaptive timing: the system applied assistance when the biological joint moment was highest and reduced or removed it when it was not needed.

A device with 24 Nm peak torque that applies torque uniformly or at the wrong gait phase can actually increase metabolic cost, because the wearer must work against the exoskeleton during phases where assistance is counterproductive. Hip exoskeleton studies have shown that phase-synchronized assistance — delivering torque during the hip extension phase of late stance — achieves meaningful metabolic reductions (IEEE TMRB, 2025), while poorly timed torque can feel like resistance rather than help.

This is why the control strategy matters as much as the torque rating. Two devices with identical peak Nm can produce opposite outcomes if one times its assistance to the gait cycle and the other does not.

Consider two devices: Device A is rated at 12 Nm peak with phase-synchronized control that delivers 10 Nm during hip extension. Device B is rated at 24 Nm peak with fixed-profile control that delivers 20 Nm during both extension and early swing. Device A reduces the wearer’s metabolic cost because all 10 Nm arrives when the body needs it. Device B may increase metabolic cost because the 20 Nm delivered during swing phase acts as resistance the wearer must overcome. The lower-torque device outperforms the higher-torque one — not because 12 Nm is inherently better, but because timing determines whether torque helps or hinders.

Hip, Knee and Ankle Torque Are Not Interchangeable

Comparing a 24 Nm hip exoskeleton to a 24 Nm knee exoskeleton is like comparing two engines with the same horsepower but mounted in vehicles with different transmissions, weights, and wheel sizes. The biological demands at each joint are fundamentally different.

Biological torque profiles differ by joint, gait phase, speed, terrain, user, and normalization method; Nm values across different joints should not be compared directly. The hip, knee, and ankle each have distinct torque demands during walking, and a device targeting one joint cannot be evaluated by the same Nm criteria as a device targeting another.

The hip joint during level walking experiences significant biological moments, particularly during the transition from stance to swing. Hip extensor and flexor moments vary with walking speed, incline, and load carriage. A hip exoskeleton providing assistance in this range can meaningfully reduce the biological hip moment, which is why most portable walking-assistance devices target the hip.

The knee joint has a more complex torque profile during walking: it absorbs energy during early stance (eccentric contraction), provides stability during mid-stance, and extends during push-off. Knee exoskeletons for squatting and injury prevention may need higher torque to handle the high knee moments during deep flexion (IEEE, 2019). But for gait assistance, the knee’s biological moment profile differs from the hip’s during level walking, meaning the same Nm rating may represent a different fraction of biological demand at the knee than at the hip.

The ankle joint produces high peak torque during push-off, but ankle exoskeletons are less common in portable consumer devices due to the difficulty of packaging actuators at the foot without interfering with ground clearance and shoe compatibility.

Some studies normalize torque to body weight rather than reporting absolute Nm, particularly in clinical populations. For example, knee extensor assistance for children with cerebral palsy has been reported as 0.09–0.38 Nm/kg during stance and 0.09–0.29 Nm/kg during swing (IEEE TNSRE). This normalization makes it possible to compare devices across different user populations, but it also means that absolute Nm ratings from different studies and different joints are not directly comparable.

Gear Ratio, Motor Power and Control Strategy

The torque rating is a product of the motor, the gear ratio, and the control system — and each of these involves engineering trade-offs that affect the wearer.

Gear ratio determines the relationship between speed and torque. A high gear ratio multiplies motor torque but reduces output speed and adds mechanical impedance — the resistance the wearer feels when moving the joint against the actuator. Quasi-direct drive (QDD) actuators use low gear ratios to preserve transparency — the ability to move freely when the motor is not actively assisting. The specific ratio depends on the motor and design goals. QDD designs sacrifice peak torque for lower impedance, which is why a QDD-based hip exoskeleton might rate at a lower peak torque while a geared design with the same motor could rate higher but feel heavier to move.

Motor power sets the energy budget. A more powerful motor can sustain higher continuous torque, but it also draws more current, which affects battery life — a critical constraint for portable devices. A 12 Nm device with a well-matched motor and battery may provide consistent assistance for a full walking session, while a 24 Nm device with an undersized battery may reduce output after 15 minutes to conserve power.

Control strategy determines how the hardware’s capability is actually used. Common approaches include:

• Proportional myoelectric control: Electromyography (EMG) signals from the wearer’s muscles drive the assistive torque in proportion to muscle activation.
• Gait-phase-based control: Pre-defined torque profiles are triggered at specific phases of the gait cycle, detected via encoders, IMUs, or foot pressure sensors.
• Adaptive/learning-based control: The system estimates the wearer’s biological joint moment and adjusts assistance in real time, as demonstrated in the Nature 2024 study.
• Model-based control: A biomechanical model of the human joint calculates the required assistance based on kinematic inputs.

Each strategy has different requirements for sensors, computation, and tuning. A device using simple threshold-based control may deliver torque at the wrong moment if the gait detection is inaccurate, while a learning-based controller may require a warm-up period to adapt to a new user.

Why More Torque Is Not Automatically More Useful

The assumption that “more Nm equals better assistance” breaks down for several reasons:

Diminishing returns on metabolic cost reduction. Studies show that metabolic cost reductions from hip exoskeleton assistance follow a diminishing-returns curve. An 18.3% net metabolic cost reduction was achieved with a portable hip exoskeleton providing up to 24 Nm peak (IEEE TMRB, 2026). Soft exoskeletons providing lower torque achieved 9.86–22.08% metabolic reductions depending on terrain (IEEE). The relationship between torque and metabolic benefit is not linear, and beyond a certain point, additional torque may not produce proportional improvement.

User comfort and acceptance. Higher torque means higher forces at the cuff-body interface, which can cause discomfort, skin irritation, or tissue compression. If the device is painful to wear, the user will not use it regardless of the torque rating. Comfort is a function of cuff design, strap geometry, pressure distribution, and the match between the assistance profile and the wearer’s gait.

Weight and battery trade-offs. Higher-torque actuators require larger motors, heavier gearboxes, and bigger batteries — or they sacrifice runtime. A 1.8 kg consumer device rated at 12 Nm (VIGX π, HKTDC product listing) is designed for all-day wearability. A 13 kg medical exoskeleton rated at higher torque (Ekso Indego, 29 lb per manufacturer listing) is designed for a different use context entirely. Comparing these by Nm alone ignores the fundamental difference in form factor and intended use.

Control mismatch. If the control strategy does not match the wearer’s gait pattern, more torque makes the mismatch worse, not better. A device that applies 24 Nm at the wrong phase creates 24 Nm of unwanted resistance. A device that applies 12 Nm at the right phase provides 12 Nm of useful assistance. The second device is more effective despite the lower rating.

What Buyers Should Ask Beyond Maximum Torque

When evaluating walking-assistance exoskeletons, distributors, research teams, and institutional buyers should ask questions that go beyond the peak Nm rating:

About torque delivery:

• What is the continuous torque rating, not just peak? At what duty cycle was it measured?
• What is the typical delivered assistance during level walking, and how was it characterized (e.g., RMS torque over a gait cycle)?
• Is the torque profile fixed or adaptive? If adaptive, what inputs drive the adaptation?

About gait-phase timing:

• How does the device detect gait phase? What sensors are used (IMU, encoder, foot pressure, EMG)?
• What is the latency between gait event detection and torque onset? How was this measured?
• Has the device been tested on the intended user population (e.g., healthy adults, elderly, patients with specific conditions)?

About fit and comfort:

• What is the cuff interface design, and how is pressure distributed?
• What is the weight of the device, including battery?
• What is the joint alignment range, and how is it adjusted for different body sizes?

About regulatory status:

• Under which regulatory framework does this device fall — personal care robot (ISO 13482), industrial robot (ISO 10218), or medical device (FDA 510(k) / EU MDR)?
• What is the intended use statement? “Walking assistance” for healthy adults is different from “gait rehabilitation” for patients with neurological conditions.
• Are there any efficacy claims? If so, are they supported by clinical evidence with defined populations and outcome measures?

About battery and runtime:

• What is the usable runtime at the typical assistance level, not at the lowest setting?
• How long does a full charge take, and is opportunity charging supported?
• What happens when the battery is depleted — does the device become transparent (free-moving) or does it add resistance?

A Torque-Spec Comparison Template

Use this template to normalize specifications across devices before making comparisons. Fill in one column per device under evaluation.

Buyer Questions Box

Before committing to an exoskeleton evaluation or pilot, confirm the following:

1. What problem are you solving? Fatigue reduction for workers, walking assistance for elderly users, or gait rehabilitation for patients? Each requires a different device class, regulatory pathway, and torque profile.
2. Who will wear it? Body size, gait pattern, and physical condition determine whether the device’s torque range and control strategy are appropriate. A device tuned for healthy adults may not work for users with gait pathology.
3. What is the intended use environment? A device for factory floor use falls under industrial safety standards. A device for hospital rehabilitation falls under medical device regulations. A device for personal walking assistance may fall under personal care robot standards. These are not interchangeable.
4. Has the device been tested on your target population? Metabolic cost reductions reported in healthy young adults do not automatically transfer to elderly users or patients. Ask for study populations, sample sizes, and outcome measures.
5. What happens when the device fails or runs out of power? Does it become transparent, lock the joint, or add resistance? This is a safety-critical question.