The Free Humanoid Corpus.
A humanoid you can print end to end but for one part — the frame, the transmission, the actuators, the skin. This is the open, CC0 prior-art package behind the printed-body topic: a single chain that runs from a readiness map of what AM can do, through a printable actuator and its family, to the cheap coupon that keeps every number honest. Every figure is derived on its own sheet or flagged as gated on a measurement.
CC0 1.0 · public domain · Physical AI × Materials × Additive Manufacturing
Six artifacts, one chain.
Each is a self-contained, interactive document, released CC0. Read them in order, or jump to the one you need — the matrix maps the problem, the four AXF-1 sheets build the actuator, and MEAS-1 and the skin budget are the measurements that gate everything above them.
AM capability matrix
The spine. Every actuator subsystem × process × demonstrated performance × TRL, with the two additive regimes — rigid/electromagnetic and soft/fluidic — and the gap between them. Nine of ten subsystems already print; the drive die is the one wall.
Open the artifact ↗AXF-1 · design target & RFQ
The printable axial-flux joint actuator, as a sheet a metal-AM bureau could quote from. Yokeless axial flux, bonded-NdFeB Halbach rotor, Cu-LPBF coils, printed cycloidal reduction — and every rating traced to the airgap-shear physics on the page, nothing fabricated.
Open the artifact ↗AXF-1 · OD sizing curve
How torque, mass, motor constant, and rotor inertia scale with outer diameter — four power laws fanning out of one baseline. The load-bearing result: there is no electromagnetic optimum; inertia and packaging set the ceiling, not magnetics.
Open the artifact ↗AXF-1 · family & humanoid budget
Four sizes (XS/S/M/L) of one design — only the diameter changes, you re-slice rather than re-engineer. The joint map, and the mass budget that decides what closes: a lean 14-DoF walker lands near 10.2 kg of actuators, 34% of a 30 kg robot.
Open the artifact ↗MEAS-1 · printed-magnet Br protocol
The one cheap coupon the whole family rides on. Measure the remanence of a printed magnet hot, at operating temperature, and convert it straight into a verdict on every joint's torque. The most likely outcome isn't "pass" — it's a quantified instruction to upsize hips or go anisotropic.
Open the artifact ↗Integrated magnetic skin · joules per taxel
The one printable perception layer where "print it" is literally true, priced in joules. The magnetized skin costs nothing to hold its field; sampling is the entire budget. Fixed-rate dense coverage is ~12 W; event-driven sampling collapses it under 2 W.
Open the artifact ↗It collapses to two problems.
Do the accounting subsystem by subsystem and the moonshot becomes a research surface. Almost every piece of a printed humanoid is already demonstrated — the whole program reduces to two things nobody has closed: putting the subsystems into one printed actuator, and the drive die, which you fab rather than print. Everything else is sizing, and every torque number in the family hangs on one cheap coupon, run twice — once for the actuator magnet, once for the soft skin.