Best Robot Actuator Torque Density: 5 Types Compared

1. Calculate rated and peak density separately

Use torque measured at the joint output. Include the motor, reducer, encoder, integrated drive and housing in assembly mass. State whether the brake, output bearing, cables and external drive are included. A motor-only ratio cannot be compared directly with a complete joint-module ratio.

The same actuator has two different ratios without any change in its hardware. A 60 Nm/kg peak value does not establish 60 Nm/kg continuous capability. Dividing torque by litres gives volumetric density; it answers a different packaging question and must use the same envelope definition for every candidate.

2. Compare architectures without inventing universal Nm/kg ranges

These categories overlap: QDD describes a drive architecture, while planetary describes a gear arrangement. Use exact models, document revisions and identical operating conditions for a numerical comparison. Neither harmonic nor planetary gearing guarantees the highest torque density, zero lost motion or a particular impact tolerance.

3. Build a comparison worksheet engineers can verify

1. Record the source: manufacturer, exact model, ratio, firmware where relevant, datasheet revision and date.
2. Normalize mass: use the complete installed assembly or explicitly list excluded parts.
3. Separate torque definitions: rated, peak, emergency and gearbox limit values are different. Record peak duration and recovery interval.
4. Match the operating point: output speed, bus voltage, ambient temperature, mounting and cooling.
5. Check the real motion cycle: acceleration, dwell, reversals and load direction. Thermal capability requires the supplier's curve or a representative test.
6. Reject incomplete comparisons: mark missing evidence as unknown rather than filling it with a category average.

For a discrete torque cycle, RMS torque is √(Σ(Torque² × duration) / total duration). It can support an initial thermal discussion, but does not by itself account for speed-dependent losses, cooling or gearbox limits. Also check the peak torque and speed at each step of the cycle.

4. Match the ratio to the actual robot joint

• Humanoid hip or knee: provide gait torque-speed traces, stance duration, peak events, joint mass budget and expected cooling.
• Robot arm shoulder or elbow: include payload, lever arm, orientation, acceleration and static holding time.
• Wrist or end effector: check package diameter, cable routing, output bearing and precision as well as torque.
• Exoskeleton joint: evaluate backdrive effort, mechanical stops, controller behavior and the intended use separately from Nm/kg.

5. Move from a comparison to a documented shortlist

First eliminate candidates that fail torque-speed, temperature, size or interface requirements. Compare mass, precision, integration effort, availability and cost only among candidates that pass. A weighted score cannot compensate for a failed engineering requirement.

Frequently asked questions

What is torque density in a robot actuator?

Torque density is output torque divided by actuator mass, usually expressed in Nm/kg. Label it rated or peak and specify the assembly mass and operating conditions.

Is harmonic torque density always higher than planetary torque density?

No universal ranking follows from the gear type. Compare exact models using the same torque definition, assembly boundary, speed, cooling and duty cycle.

Can peak density size a continuously loaded joint?

No. A continuously loaded joint must meet the required torque and speed within its thermal limits. Peak density is only meaningful with the allowed peak duration and recovery conditions.