🧪 New tutorials are being published — build from robot arms to sensors step by step
Juxi Technology Wiki

English

简体中文
繁體中文
日本語
한국어
Deutsch
Français
Español
Italiano
Português (Brasil)
Português (Portugal)

English

简体中文
繁體中文
日本語
한국어
Deutsch
Français
Español
Italiano
Português (Brasil)
Português (Portugal)

Appearance

🤗Store
/

SO-ARM101 Robotic Arm Tutorials ​

image.png

What Is Embodied Intelligence? ​

Intelligence with a body. It connects AI to various physical hardware entities, such as:

Quadruped robot dogs, bipedal humanoid robots, wheeled-legged robots, drones, self-driving cars

What Is LeRobot? ​

LeRobot is HuggingFace's open-source embodied intelligent robot software framework

GitHub address: https://github.com/huggingface/lerobot

It lowers the barrier to data collection, algorithm training and inference deployment for reinforcement learning and imitation learning (VLA), with imitation learning (VLA) being the main focus

  • Which robots can be developed with LeRobot?

From the thousand-yuan-class SO-ARM 101 robotic arm and the LeKiwi cart, up to the tens-of-thousands-yuan AgileX piper arm, the Huaxinjing StarAI arm and the Hope-JR dexterous hand, and on to the hundreds-of-thousands-yuan Unitree G1 humanoid robot. LeRobot has become the standard for data collection and algorithm training in the embodied intelligence industry.

You can also adapt your own robot to the LeRobot framework.

  • LeRobot datasets and models

LeRobot defines its own imitation-learning dataset format. You can view, use, download and train on all public datasets and models on HuggingFace, and you can also upload your own datasets to HuggingFace.

What Is the SO-ARM 101 Robotic Arm? ​

This tutorial takes the SO-ARM 101 robotic arm as an example; it uses 3D-printed structural parts and Feetech servos, at a very low cost.

This is an embodied-intelligence body even a poor student can afford, and it is one of the bodies officially recommended by LeRobot.

The arm consists of two arms: a Leader arm and a Follower arm. Each arm has 5 degrees of freedom plus 1 gripper degree of freedom.

What Computer Configuration Do I Need ​

An ordinary Windows laptop can handle everything up to training.

An ordinary Mac can handle everything.

An Ubuntu machine with an NVIDIA GPU can handle everything.

In this tutorial we use a cloud GPU platform to train models, so your own computer does not need a high-end configuration.

What Is Imitation Learning and VLA? ​

Humans drag the robot to demonstrate and collect a dataset. That dataset is then used to train an imitation-learning algorithm, which is finally deployed on the robot, letting it imitate human actions autonomously and generalize to the real environment. No teleoperation or remote control is needed.

For example, in the video above, a human drags the SO-ARM robotic arm to grab a crayfish, dip it in seasoning and drop it into hot oil, and the arm ultimately performs this action on its own. Even with a new crayfish, it can react and complete the action at any time.

Imitation learning also has a cutting-edge, fashionable name: VLA (Vision-Language-Action large model). This is also the embodied-intelligence research field that is currently developing fastest, attracting the hottest investment, seeing the fiercest China-US competition, enjoying the most prosperous open-source ecosystem, drawing intense media attention, and drawing in countless master's and doctoral students.

The algorithms LeRobot mainly adapts are imitation-learning ones, such as ACT, Diffusion Policy, SmolVLA, Pi0, Pi0.5, Wall-OSS and others.

The imitation learning in this tutorial is exclusively VLA.

Tutorial Navigation ​

Complete Courses ​

Assembly & Extension Tutorials ​

Basics ​

LeLab Tutorial ​

XLeRobot Docker Image Tutorial ​

Related Products

SO-ARM101 Developer Kit→