INDUSTRY · HUMANOID ROBOTICS

Humanoid Robotics Parts — Where Weight, Load and Cable Routing All Matter at Once

A humanoid joint is not a bracket. It carries real load, it has to be light enough for the actuator to move, and the cables and sensors have to pass through it. Additive manufacturing happens to be genuinely good at that combination — which is why this sector is the one we have chosen to specialise in.

See what we print
±0.2 mmMetal tolerance
≥99.5%Metal density
SLM + SLSPrimary processes
1 pcMinimum order
Titanium robot joint component (SLM)
OVERVIEW

Why this sector is different

A robot joint has three requirements that fight each other. It has to be stiff enough that the linkage does not flex under load, light enough that the actuator can accelerate it all day, and hollow enough to route cables, sensor leads and sometimes a cooling path through the middle. Machining gives you the first one and makes the other two expensive: you drill straight holes and then bolt plates together until it fits, and every bolted joint is mass and compliance you did not want. SLM just prints the load path you designed and leaves the rest empty.

The second reason is the iteration speed, and in robotics it is arguably the bigger one. A humanoid is redesigned constantly — a joint moves, an actuator changes, an encoder needs somewhere else to live. With a machined part each change means a new drawing, a new setup and a new lead time. With a printed part it means a new file, and the new version is in your hand in the same week the old version was. Teams that print their structure iterate two or three times more often in a year than teams that machine it, and that difference compounds.

What we would not do is print the whole arm. Bearings are bought, not printed. Gears and harmonic drives are bought or machined, because the tolerances and the surface finish they need are beyond what a printed surface delivers reliably. Sealing faces and precision bores get machined after printing. The useful division is: print the structure that is expensive to machine and cheap to print, and buy the components that are already made in volume by specialists. We will say so when a part lands on the wrong side of that line.

WHAT WE PRINT

What We Print for Robotics Teams

The parts that come through this shop repeatedly. Most of them are structural, most of them consolidate something that used to be several machined pieces, and most of them have wires running through them.

01

Joint Housings & Actuator Bodies

Load-bearing joint structure that also houses the actuator and routes its cabling — printed as one part instead of a stack of machined plates and standoffs.

02

Skeletal Frames & Links

Topology-optimised limbs and links: stiffness where the load path runs, material removed everywhere else, in titanium or aluminium.

03

End-Effectors & Gripper Jaws

Low-volume, high-variation parts that change with every payload — printed in metal for force or nylon for compliance and quick replacement.

04

Sensor & Cable Routing Brackets

Mounts and guides with the channel built into the part, so a lead is protected by structure rather than clipped to the outside of it.

05

Enclosures & Covers

Shin, torso and head shells in PA12 nylon: thin, tough, and with mounting bosses and vent paths printed in place.

06

Battery & Power Hardware

Packs, mounts and busbar hardware, with copper printed where the current path has to be shaped rather than bent.

07

Fixtures for Build & Calibration

Jigs and fixtures for the assembly cell and the calibration rig — typically SLS, made in a day, modified in an hour.

08

Tendon & Transmission Hardware

Pulleys, guides, tensioners and non-critical transmission parts, in metal where strength matters and nylon where weight does.

See the parts gallery for examples across SLM, SLA and SLS, grouped by process.

PROCESS FIT

Which Process We Would Use

Tolerance and lead time are read from the process pages, so these are the same figures the quote engine works to — not a second copy that can drift.

ProcessWhy it fits this sectorTypical toleranceTypical lead time
SLM Metal 3D PrintingLoad-bearing structure. Titanium for strength per gram and corrosion resistance, aluminium where the thermal path matters or the budget is tighter. Internal channels and consolidated geometry come free.±0.2 mm7–10 days
SLS Nylon 3D PrintingFrames, covers and brackets in PA12 nylon. No supports at all, so a complex cover prints as one piece whatever is inside it, and it takes knocks better than resin.±0.3 mm3–5 days
SLA Resin 3D PrintingAppearance and fit models, and any part with a fine feature or a smooth surface. Also the fastest route to a shape in your hand for a design review.±0.1 mm2–4 days

These are typical values we hold routinely, not absolute limits — send us the drawing and we will confirm what is achievable on your geometry. Compare all processes, or see post-processing & finishing for what happens after the print.

MATERIALS

Materials for Robotics

Four materials cover nearly all of it: two metals for structure and two nylons for everything around it.

Titanium Ti-6Al-4V

The reference SLM alloy. Highest strength-to-weight of anything we print, and effectively immune to salt water. The default for robot joints, aerospace brackets and non-implant medical instruments.

Full material data →

Aluminium AlSi10Mg

Low density, useful strength and excellent thermal conductivity. The material for weight-critical brackets, heat sinks and RF housings in automotive, drones and electronics.

Full material data →

Nylon PA12

Tough, chemically resistant and dimensionally stable, and it prints with no supports to remove. The default material for functional prototypes, brackets, housings and low-volume end-use parts.

Full material data →

Copper & Copper Alloys

Up to 100 % IACS electrical conductivity and ~400 W/m·K thermal, in three-dimensional geometry that machining cannot produce. The hardest metal to print — and quoted only after an engineer reviews the part.

Full material data →

Copper is the specialist one — printed copper has real electrical and thermal value but is the hardest of the four to build well, so send the geometry early and we will tell you if it is realistic.

QUALITY

What We Plan Before the Build

For structural parts, the decisions that matter are taken before the machine starts. These are the ones we take with you.

  • Print orientation agreed against the load path, not chosen to fit the build plate — anisotropy is real and it decides where a part can fail.
  • Support plan reviewed so witness marks land on non-functional faces, and support removal is possible from the geometry as designed.
  • Stress relief for every metal part as standard; hot isostatic pressing available for fatigue-critical components.
  • Threads, bearing seats, mating faces and sealing surfaces machined after printing, to a tolerance a printed surface cannot hold.
  • Dimensional report on request, and a first-article inspection on a pilot batch before the rest of the run is released.
  • Build parameters and material lot recorded per part, so a repeat prints the same way and can be traced if it does not.
HONEST LIMITS

Where Printing Is Not the Answer

Structural work is where an overstated claim hurts most, because the part carries load. This is our honest list.

Strengths

  • Load paths that cannot be machined: internal channels, organic structure, weight removed in the right places
  • Part consolidation — a stack of machined plates and fasteners becomes one printed body with no bolted joints to creep
  • Weight reduction without a tooling cost, which makes a redesign affordable at any point in a programme
  • Titanium and aluminium with mechanical properties in the wrought range, at ≥99.5 % relative density
  • Small quantities are economic, which matches a sector that builds tens of units rather than tens of thousands
  • One supplier for the metal structure, the nylon frames and the finishing, so interfaces are agreed in one place

Limits to know

  • Not a route to high-volume parts — for thousands of units, casting or moulding wins on cost
  • Anisotropy means a printed part is weaker across the layer direction, so it cannot be treated as isotropic stock
  • Printed surfaces cannot hold a bearing fit, a thread or a seal — those features must be machined or bought
  • Fatigue-critical parts need hot isostatic pressing and testing we do not do in-house
  • Build envelope limits size, so a large limb section may need splitting and welding
  • Printed gears, harmonic drives and bearings are not a substitute for bought components — we will say no to that
FAQ

Humanoid Robotics & Automation — Common Questions

Can you print a whole robot arm or leg?

No, and nobody should. We print the structural parts: joint housings, frames, links, brackets and covers. Bearings, gears, harmonic drives, actuators, encoders and fasteners are bought from specialists who make them better and cheaper than any printing process can. The useful split is that we make the geometry that is expensive to machine, and you buy the components that are already made in volume.

Which metal should a joint housing be printed in?

Titanium Ti-6Al-4V when strength per gram and corrosion resistance decide the design, which is most joints on a humanoid. Aluminium AlSi10Mg when the part also has to move heat, or when the budget is the constraint — it is lighter per unit volume but weaker and less stiff. Send the load case and the mass budget and we will tell you which one we would pick, and where the part is marginal.

How do you handle a bearing bore or a threaded hole?

We print a pilot and machine it afterwards. A printed surface is rough and its tolerance is an order of magnitude looser than a bearing seat needs, so printing the feature and hoping is how a joint ends up with play in it. The machining step is quoted with the print, not discovered later.

Can you print internal cable channels?

Yes, and it is one of the reasons printing wins on a joint. The constraint is that un-melted powder has to be evacuated, so a channel must be reachable from outside the part — a fully sealed internal void cannot be cleared. Design the channel with a clear escape path and we will build it.

How many units can you make, and does the price hold?

A research unit is one order; a pilot build is tens. Up to those quantities the price per part is stable, because the cost is machine time and material rather than tooling. When the volume starts to argue for casting or moulding we will say so — that is a decision you should make with the number in front of you.

Do you sign an NDA, and who sees the file?

Yes, and we would rather sign one than not. Files are treated as yours, access is limited to the people planning and running the build, and we do not show client parts in our own marketing without written permission. Humanoid geometry is competitive advantage — we treat it that way.

What do you need from us to quote?

A STEP file if you have one, plus the load case and the mass target for structural parts. If the design is still moving, send what exists and describe the constraint — we will tell you which features are printable as drawn and which will need a change before it is worth quoting.

GET STARTED

Send the Joint, the Load Case and the Mass Target

We will tell you which process we would use, where the part is marginal, and what we would machine rather than print. If it belongs with us, you get a price and a build plan; if it does not, we will say so.

Upload a File

Built for global product teams  ·  Worldwide shipping via DHL / FedEx / UPS