Skip to content
EDWartens UK

Free course · Robotics

Physical AI: Teach a Robot Arm with LeRobot

Teach a robot arm by showing it, not by programming points: imitation learning, LeRobot and its datasets on the Hugging Face Hub, PushT and ALOHA in simulation, ACT and diffusion policies, the SO-101 arm, teleoperation and recording, training and honest evaluation, SmolVLA and vision-language-action models, Isaac Sim and sim-to-real, and the safety rules that keep learned control out of safety functions.

See the modules

Free to learn. Verifiable certificate £19 when you pass the final exam (optional, VAT included).

Level
Intermediate
Modules
13
Video
18 hours
Price
Free

Outcome

What you will be able to do

Explain how a learned policy differs from a taught-point robot program; load and inspect LeRobot datasets from the Hub; explain behaviour cloning, distribution shift and action chunking; evaluate a pretrained policy in the PushT simulator and train ACT on a simulated task; choose between ACT, diffusion and SmolVLA for a task and GPU; assemble, set up and calibrate an SO-101, teleoperate it and record demonstrations; train, deploy and evaluate a policy with an honest success rate; describe sim-to-real transfer and domain randomisation; and write the risk assessment that keeps a learned policy away from safety functions.

  • Explain how a learned policy differs from a taught-point robot program, and where each belongs
  • Load LeRobot datasets from the Hugging Face Hub and read their observations and actions
  • Explain behaviour cloning, compounding error and why action chunking and diffusion help
  • Evaluate a pretrained policy in the PushT simulator and train ACT on a simulated ALOHA task, with no hardware
  • Choose between ACT, diffusion and SmolVLA for a task, a GPU and a number of demonstrations
  • Set up, calibrate and teleoperate an SO-101 arm and record clean demonstrations (optional hardware path)
  • Train, deploy and evaluate a policy with a written protocol and an honest 95 percent interval
  • Explain sim-to-real transfer and domain randomisation, and keep learned control out of safety functions under ISO 10218:2025

Software

LeRobot (open source, Apache 2.0, version 0.6 line in October 2026) with Python 3.12 and PyTorch, Google Colab for a free GPU, the PushT and ALOHA simulation environments, and the Hugging Face Hub. Isaac Sim is free but optional and needs an RTX GPU.

Hardware

None required: every module has a simulation-only path that runs on Colab. Optional: an SO-101 leader and follower arm kit with two USB cameras, and an NVIDIA GPU with 8 GB or more for local training. Isaac Sim needs a recent RTX GPU with about 16 GB of memory.

Syllabus

13 modules

Each module has video lessons, notes and a short quiz. The final exam closes the course.

  1. 1From programmed robots to learned skills: what physical AI is4 lessonsLab
  2. 2The LeRobot stack and datasets on the Hugging Face Hub2 lessons
  3. 3Imitation learning: behaviour cloning and why it drifts3 lessons
  4. 4Simulation first: gym environments and PushT on Colab2 lessons
  5. 5ACT: action chunking with transformers, explained2 lessons
  6. 6Diffusion policy and choosing a policy2 lessons
  7. 7SO-101 hardware: motors, assembly and calibration (optional)3 lessons
  8. 8Teleoperation and recording demonstrations3 lessonsLab
  9. 9Training on real data, deploying and evaluating2 lessons
  10. 10Vision-language-action models and SmolVLA4 lessons
  11. 11Isaac Sim and sim-to-real: an overview (optional, needs an RTX GPU)6 lessons
  12. 12Safety: cobot standards, guarding and where learned control is not allowed3 lessonsLab
  13. 13Final Assessment1 lesson

How it works

How this course works

  1. Watch the video lessonsOne module at a time, at your own pace.
  2. Take the module quizA short quiz at the end of each module shows what stuck.
  3. Keep your notesWrite in My notes beside each module and ask the trainer if you are stuck.
  4. Practise in the labOptionalTry a task from the module in the EDWartens Lab, in your browser.
  5. Sit the final examPass it once you have been through the modules.
  6. Certificate, if you want oneOptionalPaid, and only after you pass. Anyone can check it online.

EDWartens Lab

Practise this in the EDWartens Lab

Jog an arm with a pendant, teach points and run a pick and place cycle in the Lab's robot simulator.

Works alongside Physical AI: Teach a Robot Arm with LeRobot. The Lab is robot cell practice in your browser: no install, no hardware, free to use. Sign in as a learner and it opens with your account.

Open the free lab See the Lab
Robot cell: Teach a robot arm.

Levels to try with this course

Optional. Skip it whenever you like: it never blocks your progress or your certificate.

Why practise in a lab

Technical work is judged on what you can do: wiring, commissioning, finding a fault. Doing a task yourself teaches more than watching it a second time, and a lab lets you try it safely before you meet real equipment. It does not replace working on real hardware.

  • Practise safely, so a mistake costs nothing.
  • The lab checks your work like a site test, so you know when it is right.
  • Free, in your browser, with nothing to install.

Certificate

An optional certificate when you pass

Learning is free. Pass the final exam and you can have a certificate for £19, including VAT, with a number employers can check online. It records that you completed a free online course; it is not a regulated qualification or a CPD certificate.

  • Your name, the course, its study time and your final exam score
  • A unique certificate number and a QR code to its verification page
  • Issued by Wartens Ltd; add it to your LinkedIn profile in one click
  • Pay now or after you pass: it is issued the moment you pass the final

How the certificate works

Sample certificate for Physical AI: Teach a Robot Arm with LeRobot

Credits

Who made the lessons

The video lessons belong to their creators. EDWartens wrote the study plan, notes, quizzes and the final exam.

  • Hugging Face

    the LeRobot research presentations on ALOHA and ACT and on Diffusion Policy, the LeRobot tutorials on assembling the SO-100, recording a dataset and evaluating a policy, the LeLab introduction and the shirt-folding recap in the final module

  • NVIDIA

    how robots learn through training, simulation and real-world deployment, the next wave of physical AI, and narrowing the sim-to-real gap with Isaac Sim

  • IBM Technology

    the explainer on what physical AI is and how robots learn and adapt

  • Google for Developers

    the talk introducing LeRobot and why it lowers the entry barrier to AI for robotics

  • RAIL

    Sergey Levine's UC Berkeley CS 285 lecture on imitation learning, part 1

  • e-Yantra

    the IIT Bombay series on the LeRobot SO-101 arm: hardware and degrees of freedom, leader-follower teleoperation, and an introduction to imitation learning

  • Welch Labs

    how vision-language-action models work inside

  • Trelis Research

    training an ACT policy for the SO-101 with LeRobot end to end

  • Articulated Robotics

    Isaac Sim in under half an hour

  • exida

    hazard analysis and risk assessment of collaborative robots

  • Ilia

    running AI robotics experiments at home with LeRobot and the SO-ARM100

  • Phospho AI

    repairing, merging and splitting LeRobot datasets

  • Sanjiban Choudhury

    the core concepts of imitation learning

  • Aleksandar Haber PhD

    the introduction to the Gymnasium interface with the Cart-Pole environment

  • Foundation Models For Robotics

    training a behaviour-cloning policy on PushT with LeRobot, SmolVLA explained, and LeRobot asynchronous inference

  • Trossen Robotics

    the ALOHA tutorial with Hugging Face LeRobot

  • Robraintics

    why diffusion policy is changing robot learning

  • Trushant Adeshara

    SO-101 motor configuration and calibration

  • Pius Lim

    ACT against SmolVLA on the same SO-101

  • Shane Reetz

    the Sim-to-Real with NVIDIA Isaac series: building a home robot lab and an SO-101 workspace

  • Lightwheel

    teleoperating the SO-101 in Isaac Sim with LeRobot, step by step

  • Learning Automation

    summaries of ISO 10218-1:2025 and ISO 10218-2:2025

Questions

About this course

Who is this physical AI and LeRobot course for?
It is for robotics, mechatronics and automation engineers and engineering students who want to move from programming robots point by point to teaching them from demonstrations. Robot programmers from FANUC, ABB, KUKA or Universal Robots backgrounds, and ROS 2 users, will find it a natural next step.
Do I need a robot arm to take the course?
No. Every module has a simulation-only path: the PushT and ALOHA simulators run on a free Google Colab GPU, and real-robot datasets, including SO-101 ones, are on the Hugging Face Hub to train on. The SO-101 modules are optional and show what to do if you buy or build an arm later.
What do I need to know first?
You should be able to run a Python notebook and follow a command line. The course explains the machine learning it uses. If Python is new, take Python for AI and Engineering Data first; for classical robot software, ROS 2 Robot Programming with Python complements this course.
Which software does it use?
LeRobot, Hugging Face's open-source robot learning library (version 0.6 line in October 2026, Python 3.12 or later), with PyTorch, Google Colab and the Hugging Face Hub. Isaac Sim appears in one optional module and needs an RTX GPU. LeRobot changes quickly, so the notes give the current commands and tell you to check the release notes.
Is the course free to learn?
Yes. Every module, the notes, the worked problems, the project and the final assessment are free, and all the software used is free. The certificate is optional.
How long does the course take?
About 17 to 18 hours of video, notes and practice at your own pace, of which about 10.5 hours is video. The optional PushT project takes about 16 hours more, most of it waiting for training runs.
Is a learned policy safe to use on a factory robot?
Only for task decisions, inside the robot's rated safety configuration and a proper risk assessment. A learned policy is never allowed to be a safety function. The safety module covers ISO 10218-1 and -2 (2025 editions), why protective stops and speed limits must be rated hardware, and how to run a first test safely.
What certificate do I get?
An EDWartens certificate of completion, issued when you finish the modules and pass the 15-question final at 60 percent, with a number anyone can verify on our site. It is not a certification from Hugging Face, NVIDIA or any robot maker, none of whom is affiliated with EDWartens.

The Professional Module is five live days of Siemens PLC, HMI and SCADA with an instructor, then a resume session and a CPD certificate.

See the Professional Module