zero2robot: interactive textbook
Skip to contentInteractive textbook<br>Build a robot brain.No robot required.<br>You start from a blank simulation loop and build up, all the way to vision-language-action policies, your own physics engine, and a real robot arm. No framework to excavate, no black boxes. Free on a Colab T4 or your laptop.<br>Train your first policyA few minutes, free.<br>No account, no setup, and your place is remembered locally in your browser.
See it work<br>live
Open chapter →‹›
4096 robots learn at once<br>ch2.3<br>You write parallel PPO on MJX. Here a 48-robot sample of the real run flails, then the whole field balances.
Now showing: 4096 robots learn at once. You write parallel PPO on MJX. Here a 48-robot sample of the real run flails, then the whole field balances.
Quantize -> BPE)"],"number":[0,"5.5"],"phaseKey":[0,"phase5_practitioner"],"phaseLabel":[0,"Phase 5 · Practitioner"],"exerciseIds":[1,[[0,"ex1"],[0,"ex3"]]]}],[0,{"slug":[0,"ch5.6-lora"],"title":[0,"LoRA From Scratch: Adapt a Frozen Policy"],"number":[0,"5.6"],"phaseKey":[0,"phase5_practitioner"],"phaseLabel":[0,"Phase 5 · Practitioner"],"exerciseIds":[1,[[0,"ex1"],[0,"ex2"]]]}],[0,{"slug":[0,"ch5.7-quantize"],"title":[0,"Quantize a Policy by Hand: INT8 Is a Scale, Not a Rounding"],"number":[0,"5.7"],"phaseKey":[0,"phase5_practitioner"],"phaseLabel":[0,"Phase 5 · Practitioner"],"exerciseIds":[1,[]]}],[0,{"slug":[0,"ch5.8-real_loop"],"title":[0,"The Real Loop: Teleoperate, Record, Train, Deploy, on a Real Arm's Body (in sim)"],"number":[0,"5.8"],"phaseKey":[0,"phase5_practitioner"],"phaseLabel":[0,"Phase 5 · Practitioner"],"exerciseIds":[1,[[0,"ex1"],[0,"ex2"]]]}]]]}" ssr client="visible" opts="{"name":"ProgressOverview","value":true}">Why this course is built the way it is
the code is the product<br>One file you can read top to bottom<br>Every chapter builds a single runnable script: a few hundred lines, no framework to excavate. You read it like a textbook, and every line you read is a line that runs.
runs on the free tier<br>A laptop or a free Colab T4 is enough<br>Every learner-facing path completes on CPU or a free T4. The wall-clock times on each page are measured on real hardware, never estimated; where a tier hasn't been measured yet, the page says so.
honest by construction<br>Real sims, real policies, real numbers<br>The browser demos run the same code you train locally. Seeds are mandatory, results reproduce within a recorded band, and every number you read traces back to a run, including the ones that show a method failing.
Phase 0 · Foundations<br>6 live<br>Make the simulator behave. Step physics by hand, author a scene, get frames and rotations right, and record your first teleoperated dataset.
Phase 1 · Imitation<br>9 live<br>Teach a policy from demonstrations. Behavior cloning, exactly why it breaks, and the models built to stop it from breaking.
Phase 2 · Reinforcement<br>8 live<br>Learn from reward instead of examples: for when demonstrations run out and the robot has to try, fail, and improve.
Phase 3 · Depth (optional)<br>9 live<br>Off the main line, taken when the itch strikes. Learn a world model, build a physics engine from scratch (dynamics, constraints, contact), compare it against the simulator you have trusted all along, then plan through your own engine with sampling-based MPC. The main line needs none of it.
Phase 4 · Post-Training<br>3 live<br>Take an already-trained policy and make it reliable. An offline-RL primer that learns from logged data, DAgger corrections that close the gaps demonstrations left, and human-in-the-loop RL (HIL-SERL) that turns a decent policy into a dependable one.
Phase 5 · Practitioner<br>8 live<br>Build the practitioner's stack from scratch. A ViT and contrastive vision-language for perception, the two-tower VLA shape and a FAST action tokenizer, LoRA and INT8 quantization by hand, then the real-arm teleop → record → train → deploy loop that graduates you off the simulator.
See the full map: every chapter, and how they connect →
Physical AI Weekly<br>The field, explained from first principles.<br>One mechanism teardown a week, plus a tight digest of what shipped. Honest, no hype.
Email address<br>SubscribeFree. No spam. Unsubscribe anytime.