Boston Dynamics Atlas electric platform note
Published April 17, 2024
Confirms the humanoid platform direction toward electric actuation and whole-body mobility requirements.
Open sourceA practical fit checker and buyer guide for selecting a humanoid robot actuator by joint, torque, mass, motion architecture, and evidence quality.
Tool
Start with a joint, torque target, mass budget, output type, and integration need. The checker returns a first actuator lane and the evidence to ask for next.
Live density preview: 32.1 Nm/kg
Result
Enter joint-level assumptions to see whether a rotary, linear, QDD, or integrated actuator path is the better first RFQ lane for actuator for humanoid robots sourcing.
First RFQ lane
Awaiting inputs
For procurement, a humanoid robot actuator is not only a motor. It is the motion stack that can be installed into a humanoid joint: motor, reducer or screw, encoder, brake when needed, driver, harness, thermal path, mechanical interface, and validation evidence.
The tool gives a first lane. The table below explains what each lane is good for, what evidence to request, and where buyers get surprised during prototype or pilot builds.
| Actuator lane | Best fit | Evidence to request | Procurement caution |
|---|---|---|---|
| Integrated rotary joint module | Hip, knee, ankle, shoulder, elbow, torso | continuous torque, peak torque duration, reducer backlash, brake option, encoder resolution, thermal mounting conditions | Peak torque can look attractive while continuous torque, heat path, and harness routing still fail the robot duty cycle. |
| QDD or low-ratio backdrivable module | Force-control joints, research platforms, compliant contact | reflected inertia, torque sensor or current-control evidence, gearbox ratio, impact tolerance, controller bandwidth | Without a backdrivable path, walking impacts (2-3× body weight) force the gearbox to absorb 100% of shock energy, leading to shear failure. |
| Frameless torque motor stack | Teams with their own reducer, housing, cooling, and electronics | stator/rotor drawing, winding option, magnet grade, torque constant, insulation class, encoder compatibility | This lane shifts integration work from the supplier module to the robot mechanical and electronics team. |
| Linear or roller-screw actuator | Hands, fingers, knees with linkages, compact stroke mechanisms | stroke, screw efficiency, life estimate, backlash, radial load, sealing, limit detection | Linear actuation is not automatically lighter once linkages, guides, sensors, and packaging are counted. |
Time-sensitive information was reviewed on July 24, 2026. The sources do not replace supplier test data, but they explain which actuator claims should be verified before samples are approved.
Published April 17, 2024
Confirms the humanoid platform direction toward electric actuation and whole-body mobility requirements.
Open sourcePublished 2021; reviewed July 24, 2026
Supports the QDD lane for low reflected inertia, torque control, and backdrivable joint behavior.
Open source2026 edition; reviewed July 24, 2026
Flags magnet-material supply exposure for actuator programs that depend on high-energy rare-earth magnets.
Open sourceCatalog PDF; reviewed July 24, 2026
Shows why torque-speed curves and continuous operating regions need to be requested before sample approval.
Open sourceCourse note; reviewed July 24, 2026
Provides the basic speed-torque framing behind continuous torque, peak torque, and thermal window questions.
Open sourceReviewed July 24, 2026
Reveals that actuators account for 30% to 60% of the total Bill of Materials (BOM), making them the primary target for cost-down engineering.
Open sourceReviewed July 24, 2026
Notes that walking generates 5,000 steps per hour with 2-3× body weight shock forces, requiring backdrivability to avoid immediate gearbox shear failure.
Open sourceReviewed July 24, 2026
Shows how maintaining an 85°C winding temperature envelope is critical to sustaining continuous plateau torque during 2-5 minute heavy duty cycles.
Open sourceMarket data and operational physics reveal that humanoid robot actuators fail or blow budgets because of three overlooked realities. These notes were reviewed on July 24, 2026, and should be confirmed against supplier test data before sample approval.
Depending on hand complexity and degree of freedom (DOF) counts, joint actuators consume between 30% and 60% of a humanoid robot’s total Bill of Materials (BOM). Moving from prototype to production requires standardizing these modules rather than custom-designing every joint.
A specification sheet peak torque is useless if it causes a thermal shutdown in 10 seconds. Humanoid operations require sustaining a plateau torque for 2–5 minute duty cycles without derating. This typically requires high Km motors capable of staying within an 85°C winding temperature envelope.
A humanoid robot takes roughly 5,000 steps per hour. Each step sends a shock of 2–3× body weight through the leg actuators. If an actuator is mechanically self-locking, the gearbox absorbs 100% of this shock, risking immediate shear failure. Backdrivability is mandatory for walking legs.
The checker screens the first architecture path. It cannot certify final actuator life, impact tolerance, sealing, controller stability, or robot-level safety. Those need sample tests and supplier data.
| Risk | Evidence | Action |
|---|---|---|
| Thermal overpromise | Torque-speed curve, continuous operating region, winding temperature limit, mounting plate condition | Request derating data before accepting a peak-torque-driven recommendation. |
| Wrong actuator lane | Joint zone, output type, backdrivability need, mass budget, cable path | Use the fit checker and compare rotary, QDD, frameless, and linear lanes before RFQ freeze. |
| Integration delay | Interface drawing, firmware protocol, encoder format, brake wiring, connector revision | Keep the first sample close to a catalog baseline and push custom work after validation. |
| Supply exposure | Magnet grade, reducer sourcing, encoder availability, controller component lead time | Ask for alternates and substitution rules when moving from prototype to pilot quantity. |
A useful RFQ for actuator for humanoid robots sourcing should make the supplier prove fit against the joint duty, not only against a model name.
joint zone, continuous torque, peak torque, output speed, duty cycle, impact or holding events
outer diameter, axial length, mass target, flange, output shaft, hollow shaft, cable exit, drawing revision
bus voltage, controller location, protocol, encoder format, brake logic, connector, harness length
Yes. On this site, actuator for humanoid robots and actuator for humanoid robot are treated as aliases of humanoid robot actuator. The canonical page is this URL, so buyers do not need a separate route for the same sourcing intent.
Ask for continuous torque with the test temperature, mounting condition, voltage, speed point, and duration. Peak torque is useful only after the continuous window is credible.
Choose a QDD or low-ratio path when backdrivability, torque control, and compliant contact matter more than compact holding torque. Ask for reflected inertia and control evidence.
Linear actuators are strongest when stroke, linkage geometry, hand packaging, or high-force compact movement is more important than a coaxial rotary output. However, self-locking linear screws in legs face high risk of shock-induced failure.
Actuators typically account for 30% to 60% of the total Bill of Materials (BOM), depending on the degrees of freedom and hand complexity. A full-size bipedal robot often needs 30-40 actuators, making joint module cost the biggest scaling bottleneck.
Include joint zone, torque-speed duty cycle, mass envelope, output interface, bus voltage, protocol, encoder, brake need, harness exit, drawing revision, quantity, and target validation date.
Usually no. Hip, knee, ankle, shoulder, wrist, hand, torso, and neck joints have different torque, speed, mass, wiring, brake, and impact constraints. Use one architecture family where possible, but qualify each joint zone separately.
A torque number is weak when it lacks speed point, duration, winding temperature, ambient condition, mounting plate, bus voltage, and cooling assumptions. Ask for the continuous operating region, not only one headline value.
Keep the first sample close to catalog form when the team still needs to validate torque-speed behavior, thermal margin, encoder quality, backlash, brake logic, or control stability. Move custom interfaces after baseline evidence is clear.
For hip, knee, ankle, and foot joints, request impact tolerance, reducer backlash, continuous torque derating, sealing assumptions, brake behavior, cable exit constraints, and backdrivability or force-control evidence.
Mark the value as provisional, ask for test conditions, and run a sample plan before freezing the robot-level design. If public evidence is thin, compare multiple suppliers and request controlled drawings plus torque-speed curves.
Send the joint map, target torque-speed window, mass envelope, voltage, output type, and intended duty cycle to suppliers, then ask for a short sample review instead of committing to a custom module immediately.
Send the checker result, joint map, envelope drawing, and duty cycle assumptions. The fastest route is usually one catalog sample, one evidence review, then custom interface decisions.
Inquiry Email
Subject: Humanoid robot actuator fit check and sample RFQ
Attach the joint map, fit-check result, interface drawing, duty cycle, and target sample date.
Instant Chat
+8618857971991
Best for quick model-fit questions before a full RFQ email.