Introducing · OpenArm 2.0

OpenArm 2.0

Physical AI, made easy for everyone.

A 100% open-source bimanual humanoid arm. Compact gripper, in-hand camera, and a variable-controlled Cell — supplied one-stop from Korea by Libertron.

Buy now →Get in touch
Order now — September delivery, limited first batch
7-DOF ×2
633mm
4.1/6.0kg
1kHz CAN-FD
A globally open-source robotics project
Works with your standard stack
ROS 2MuJoCoIsaac SimPythonCAN-FDOpen-source CAD·BOM
0-DOF ×2
Per dual arm
0mm
Reach (CAD-measured)
0/ 6.0kg
Nominal / peak payload
0kHz
CAN-FD control
0%
Open-source CAD & firmware
Core

Why you should buy OpenArm 2.0

We refined what fell short in 1.0 — reborn as a next-generation platform where reinforcement and imitation learning run the way they should.

A smaller gripper, an in-hand camera

Simpler actuation makes the gripper smaller. The in-hand camera captures the moment of grasping, and fingers swap to fit each task. Upgraded from a chest-mounted camera to a head-mounted ZED stereo camera, providing a more natural top-down view and depth-rich training data.

// IN-HAND CAMERA · swappable fingers · ZED recommended

Motorless KER add-on

Collect and feed imitation-learning data faster and more easily with KER, a lightweight motorless input device.

Intuitive teleoperation

Smooth gravity compensation lets you operate precisely, as if demonstrating the task by hand.

Fully open source

CAD, firmware, and control code are all open — ROS2-compatible for rapid development.

Ships fully assembled

Libertron delivers OpenArm 2.0 ready to use the moment it arrives — assembly and setup already done.

Hardware

Spin the 2.0 hardware yourself

Drag to explore in 360° — from the body with its in-hand camera to the calibration jig that keeps your dataset consistent.

// UNIT_2.0 + HEAD_CAMDrag · 360°

OpenArm 2.0 · head camera

The full setup — the body with integrated in-hand camera plus the top-down camera.

// CALIBRATION_JIGDrag · 360°

Calibration jig (CELL only)

When operating OpenArm 2.0 in the CELL, it locks the gripper to its exact CAD-defined angles — calibrating out assembly errors for a consistent dataset.

◇ Inside

2.0's redesigned gripper

A built-in camera — and a gripper you can 3D-print and swap yourself.

drag to rotate · scroll to explode
drag to rotate · scroll to explode
◆ Applications

Use it for research like this

From teleoperation data collection to reinforcement and imitation learning, humanoid research, and education — one OpenArm opens the door to a wide range of physical-AI work.

OpenArm use case: Teleoperation & demonstration learning
Teleoperation & demonstration learningCapture human demonstrations directly via leader–follower teleop.
OpenArm use case: Reinforcement & imitation learning
Reinforcement & imitation learningTrain policies on the collected trajectories, then validate on hardware.
OpenArm use case: Robot manipulation research
Robot manipulation researchDevelop precise bimanual grasping, assembly and handover algorithms.
OpenArm use case: AI-driven humanoid learning
AI-driven humanoid learningPort bimanual manipulation data into humanoid skills.
OpenArm use case: Human–robot interaction (HRI)
Human–robot interaction (HRI)Study safe, force-controlled close collaboration and contact.
OpenArm use case: AI robotics education
AI robotics educationA hands-on robotics curriculum on the ROS2, Python & CAN-FD API.
Ecosystem

Make 2.0 more useful — powerful add-ons

// MOD_KEROpenArm KER
Coming soon · CAD·BOM later

OpenArm KER

A motorless leader arm with the exact same joint structure as 2.0 (Kinematic Equivalent Replica). No motors means it's light and fatigue-free — ideal for teleoperation and teaching-data collection.

Identical structure & designSame joints as 2.0 for 1:1 mapping
MotorlessLight and affordable without motors
Drag-free operationA comfortable weight even in long demos
Data collectionIdeal for teleop and teaching
// MOD_CELL
Coming soon

OpenArm Cell

A reproducible evaluation cell that keeps background, lighting, cameras, and arm position identical every time — a standard environment for fair, automated model comparison.

Standard environmentControls background, lighting, cameras for fair comparison
Z-axis liftHeight adjust to reproduce varied tasks
Safe & long-runningMotion-detection sensors for safer testing
Automated evaluationReproducible setup for automatic model comparison
▶ Reveal

The moment it was first revealed

The official video that introduced OpenArm to the world — see, at a glance, the concept of an arm built to make physical-AI research affordable.

Libertron

Why buy from Libertron?

OpenArm is an open-source project available worldwide. Libertron's role is simple — we eliminate the 3+ months and countless trials you'd otherwise waste on sourcing, assembly, and setup. Now, simply focus on your research.

01

“Zero assembly”

Delivered fully built

Assembled, calibrated, and tested by Libertron engineers before shipping.

DIY 180+ hours→0 hours
02

“Ready on arrival”

No setup period before you start

Fast single-vendor delivery with complete documentation, all in one place.

DIY 2–3 months→Start immediately
03

“Pre-tested software”

Run it the day it arrives

Basic operation verified, with ready-to-run example code included.

Self troubleshooting→Example code included
04

“Technical support”

Maintenance is on Libertron

Transparent, fast B2B technical support.

100% on you→Libertron support
🤝 Libertron Devkits

When you factor in time and risk, it's ultimately the most affordable choice.

Get a quote in the store →
Partnering with Libertron

End-to-end support

From sourcing and customs to assembly, calibration, and first run — we hand it over ready for research to begin.

A platform that keeps improving

Just as 1.0's field experience shaped 2.0, we keep updating firmware, example code, and know-how alongside you.

An expandable solution

KER, Cell, camera packages — scale your entire research setup and get quotes from a single point of contact.

Resources

Code, docs, and a community — all open

OpenArm is an open-source project anyone can inspect and build on. From hardware CAD to firmware and control code to a worldwide developer community — explore it, use it, and help it grow.

GitHub repository

Hardware CAD, firmware, and control code are all public. Clone it, build it, and modify it freely.

View code →

Official docs

API reference, setup guides, and tutorials. (English)

Read docs →

Discord community

Ask, answer, and build together with developers around the world in real time.

Join server →
Browse all resources →
? FAQ

Frequently Asked Questions

Here are the most common questions about the OpenArm robotic arm.

How is OpenArm different from existing industrial robotic arms?
OpenArm is the ultimate starting machine to experience collaborative robots at a reasonable price. Unlike expensive industrial robots, we eliminated price bubbles while keeping research-grade specifications. It is comprehensively designed based on open-source so anyone can easily access and modify it, making it optimized for education, research, and prototyping.
Can I control it directly using Python or ROS2?
Yes, absolutely! OpenArm provides a Python API and can be directly coded and controlled in various environments including C++, ROS2, and WebSockets. It is the best platform for robotics students or AI researchers who want to quickly deploy deep learning and reinforcement learning (RL) models onto a physical robot arm.
Is it suitable for university labs or high school robotics courses?
It is the best choice. With its compact size and light weight, you can safely experiment even in tight spaces. Students teach it intuitively — guiding the arm by hand (kinesthetic teaching) or demonstrating with the leader arm — and its open-source nature supports everything from basic coding education to advanced robotics dynamics research.
Can I attach accessories like grippers or cameras?
Yes. Beyond the standard Leader and Follower grippers, you can test custom grippers, suction pumps, and other end effectors. Because the hardware is open source, fabricating and mounting your own adapters with a CNC or 3D printer is straightforward. A depth camera (ZED Mini stereo camera) that captures depth data can be mounted on top, so you can also try vision-AI-based autonomous pick-and-place.
How does after-sales and technical support work?
Libertron provides a one-stop solution covering the entire process of equipment adoption from consultation and delivery to technical support based on products manufactured in Korea. For customers in Korea, we ensure fast and seamless supply without complex import procedures. For international customers, products are delivered safely and efficiently through global logistics partners such as FedEx.
Purchase

Meet OpenArm 2.0 — available now

Add the configuration you need and order, or request a quote.

Buy now → View OpenArm 1.1
Order now — September delivery, limited first batch
Contact

Send us your questions

Leave an inquiry or quote request and our team will get back to you shortly.

TEL +82 (02) 3486-5278
MAIL openarm@libertron.com
ADDR Room 1111, SK V1 Center W, 11 Dangsan-ro 41-gil, Yeongdeungpo-gu, Seoul