Modern Robot Learning from Scratch - Lecture 6 - Train ACT Model on SO-101 Robot from Sratch

Every field has a few ideas that quietly change everything. For large language models, it was the Transformer. For modern robot learning, I would argue it is the ACT policy (Action Chunking with Transformers). The ACT policy fundamentally changed what was possible in robotics by opening the door to open-source, learnable policies that could actually run on low-cost hardware. That single shift moved robot learning out of a few elite labs and into the hands of much smaller research teams. Robots trained with ACT could fold clothes, insert batteries, close zip-lock bags, and handle a wide range of real-world manipulation tasks that were previously brittle or hand-engineered. This lecture teaches you how to implement the ACT policy in the SO-101 Robot from Scratch. I walk you through all the steps, even the parts where I got stuck and how I overcame it. This lecture is meant for beginners to write their first AI policy in an hardware and make it work. This lecture is like the ACT series finale! We learnt all about the ACT policy so that we could know what we are deploying in the Robot! These are the three links which I use for most of the video: Our GitHub Repo: https://github.com/VizuaraAI/Modern-Robot-Learning LeRobot Installation instructions page: https://huggingface.co/docs/lerobot/en/installation Setting up SO-101 with LeRobot: https://huggingface.co/docs/lerobot/en/so101 WowRobo Website: https://shop.wowrobo.com/products/so-arm101-diy-kit-assembled-version-1 Document which I refer to many times! (contains all instructions which I used)- https://docs.google.com/document/d/1aKPEzLnidv257Ud4zU5Scn7rtATzx4U1-ng4vUZeqnw/edit?usp=sharing We would be happy if people get inspired by this to get into the field of Robotics and try this out!