How Asimov 1 is put together

One robot, five layers

Layer	What it is	Where it's published
Structure	7075 aluminium, MJF PA12 nylon, 316L sheet metal, bought-in parts	mechanical/ASV1 (STEP), fabrication manifest
Actuation	25 ENCOS actuators in 5 families, 48 V, CAN	docs.menlo.ai mechanical tour
Electronics	RPU (MCB + PDB), head stack, 13S4P battery, harness	electrical/ (KiCad, EasyEDA, cable CSVs)
Software	RPU and Pi 5 images, Asimov Manager, firmware	docs.menlo.ai setup
Learning	MuJoCo model, PPO + AMP walking policy, training code	sim-model/, Hugging Face, cyclotron

Menlo's published figures: 1.2 m, 35 kg, 25 powered degrees of freedom. The simulation model in this course stands 1.25 m and its links add up to about 32 kg.

What this course adds to the introduction

The introduction taught you to identify, connect and configure each part. Here you go below that:

• Design: why the legs are built the way they are, how actuators are sized, how the PCBs are laid out, and how the battery is protected.
• Build: what is machined, bent, printed or bought; what the 165 manufactured and bought parts are; how to source them.
• Control and learning: the PD control loop, the reinforcement-learning recipe Menlo trained the walking policy with, and running that real policy (and your own) in the Training Bay.

You'll finish by wiring and repairing the harness in the STEM Engine, training and uploading a policy, and walking the Interstellar Hub as your own Asimov.

---

Asimov 1 by Menlo Research · CERN-OHL-S-2.0. Sources: the asimov-1 repository, docs.menlo.ai, and Menlo Research's training files (BSD-3).