UFACTORY xArm Series

Home > Hardware > Arms > xArm

Note: Prices and specifications may vary by region and time. Please refer to the UFACTORY official site for the latest information.

Overview

Item	Details
Manufacturer	UFACTORY (Shenzhen, China)
Year Founded	2013 ¹
Type	Collaborative Robot
Features	Cost-effectiveness, Research accessibility, Open-source SDK
Global Sales	80+ countries (per official announcement) ¹

Company Introduction

UFACTORY is a robot manufacturing startup founded in 2013 in Shenzhen, China ¹. They gained recognition by launching uArm, the world’s first desktop open-source robot arm, on Kickstarter, and the uArm Swift PRO raised over $1 million in crowdfunding platforms ².

Their core strategy is to leverage Shenzhen’s manufacturing ecosystem to offer industrial-grade technology at accessible prices for the research/education market. Investors include Tencent AI Accelerator and Sinovation Ventures ².

Key Significance

Democratization of Research Accessibility

The xArm series is a product line that significantly lowered the barrier to entry for robot research:

Price Competitiveness: Provides similar specifications at approximately 1/2 to 1/3 the price of existing research robots ³
Research-Grade Precision: Capable of performing research-level manipulation tasks based on repeatability specifications
7-Axis Option: xArm 7 offers flexibility similar to the human arm with 7 degrees of freedom
Complete SDK Support: Official Python, C++, ROS/ROS2 support ⁴

Comparison with Franka

Item	Franka Research 3	xArm 7
Price Range	High (research institution quote)	Mid-low (~$10,000s) ³
Payload	3kg ⁵	3.5kg ³
Repeatability	0.1mm ⁵	0.1mm ³
Degrees of Freedom	7 DoF	7 DoF
Torque Sensing	Built into all joints ⁵	Optional (external sensor)
ROS Support	ROS, ROS 2	ROS, ROS 2 ⁴

Note: Franka Emika filed for insolvency in 2023 and was acquired by Agile Robots. Currently operating as Franka Robotics and support status may change. Check official channels for latest information.

Franka has advantages in hardware features like all-joint torque sensing, but xArm is being used as a research alternative due to accessible pricing and stable support.

Product Lineup

Specifications below are based on UFACTORY official materials. For accurate specifications, please refer to the official product page.

Model	Axes	Payload	Reach	Repeatability	Max Speed	Weight	Price (USD)
Lite 6	6	600g	440mm	0.5mm	500mm/s	~7kg	$2,999~ ⁶
xArm 5	5	3kg	700mm	0.1mm	330mm/s	~11kg	$5,000~ ³
xArm 6	6	5kg	700mm	0.1mm	1m/s	~12kg	$8,000~ ³
xArm 7	7	3.5kg	700mm	0.1mm	1m/s	~14kg	$10,000~ ³
UFACTORY 850	6	5kg	850mm	0.02mm	1m/s	~20kg	$8,999~ ⁷

Model-Specific Features

Lite 6

Entry/Education: Start robot research at the lowest price
Lightweight Design: Easy to move and install
Integrated Control Box: Space-saving design
Suitable For: Lab automation, education, touchscreen testing

xArm 5

5-Axis Entry Model: Entry version of industrial robots
3kg Payload: Can perform most research manipulation tasks
Value: Most economical full-size model in the series

xArm 6

6-Axis Standard Model: Most versatile choice
5kg Payload: Highest payload in the series
1m/s Speed: Fast task execution

xArm 7

7-Axis Premium Model: Human arm-like degrees of freedom
Redundant Joint: Advantageous for obstacle avoidance, singularity avoidance
AI/CV Research Optimized: Ideal for VLA model training

UFACTORY 850

Latest Flagship Model: Extended workspace
Extended Reach: 850mm for wider work area
Highest Precision: Top level in the series (see specifications table above) ⁷
Industrial-Grade Harmonic Drive: 24/7 operation design
DC/AC Options: Supports both mobile platforms (DC) and fixed installation (AC)

Common Features

Carbon Fiber + Aluminum Structure: Lightweight design
Industrial Harmonic Drives: Durability and precision
Multi-turn Absolute Encoders: Precise position control
Web-based UFACTORY Studio: Windows, macOS, Ubuntu support
Blockly-based Visual Programming: Control robots without coding

Software and SDK

Official SDKs

SDK	Description	Link
xArm-Python-SDK	Python-based control API	GitHub
xArm-CPLUS-SDK	C++-based control API	GitHub
xarm_ros	ROS 1 package	GitHub
xarm_ros2	ROS 2 package (Gazebo, MoveIt support)	GitHub

Code Example

from xarm.wrapper import XArmAPI

arm = None  # Initialize for cleanup on connection failure

try:
    # Connect to robot (modify IP address for your environment)
    arm = XArmAPI('192.168.1.113')  # or serial: 'COM5'

    # Initialize
    arm.motion_enable(enable=True)
    arm.set_mode(0)
    arm.set_state(0)

    # Move to position (units: mm, degrees)
    arm.set_position(x=300, y=0, z=200, roll=180, pitch=0, yaw=0)

except Exception as e:
    print(f"Error: {e}")
    if arm and arm.connected:
        arm.emergency_stop()  # Emergency stop

finally:
    # Cleanup: only perform if connected
    if arm and arm.connected:
        arm.motion_enable(enable=False)
        arm.disconnect()

Caution: When controlling real robots, be familiar with safety procedures and always keep the emergency stop button accessible. See UFACTORY documentation for detailed safety guides.

ROS 2 Support Features

MoveIt 2 integration
Gazebo simulation
Velocity control mode (firmware 1.6.8+)
RealSense D435i integration demo

VLA Research Applications

Open X-Embodiment Dataset

xArm is one of the major robot platforms in the Open X-Embodiment dataset ⁸:

One of 22 robot platforms in the dataset (alongside Franka, WidowX, Google Robot, etc.)
Includes various forms such as single-arm manipulators, dual-arm robots, and mobile platforms
Total 1M+ trajectories, 527 manipulation skills (21 institutions participating) ⁸

Source: Based on Open X-Embodiment paper (arXiv:2310.08864). Dataset is continuously updated, so see official project page for latest figures.

RT-X and VLA Models

Major models trained on Open X-Embodiment data:

Model	Description
RT-1-X	Cross-embodiment generalist policy
RT-2-X	Vision-Language-Action model
OpenVLA	7B parameter open-source VLA ⁹

Diverse robot data including xArm is used for cross-embodiment generalization research in these models.

LeRobot Framework

Hugging Face’s LeRobot project provides xArm-related datasets ¹⁰:

Pre-trained Models: Diffusion Policy, ACT, VQ-BeT, etc.
Standardized Data Loaders: Supports various robot formats
Hugging Face Hub Datasets:
- lerobot/xarm_lift_medium
- lerobot/xarm_push_medium_replay_image

# Loading xArm dataset in LeRobot
from datasets import load_dataset

dataset = load_dataset("lerobot/xarm_lift_medium")
print(dataset)

Research Use Cases

Imitation Learning: Demonstration data collection and policy learning
Sim-to-Real Transfer: Transfer from Gazebo simulation to real robot
Multi-task Learning: Learning various manipulation tasks
Foundation Model Fine-tuning: Fine-tuning pre-trained models like OpenVLA

Accessories and Peripherals

Accessory	Purpose
xArm Gripper	Standard 2-finger gripper
xArm Vacuum Gripper	Vacuum suction gripper
BIO Gripper	Flexible gripping
6-Axis Force Torque Sensor	Force/torque feedback
Linear Motor	Linear extension rail

Purchase Guide

Recommendations by Research Purpose

Purpose	Recommended Model	Reason
Education/Entry	Lite 6	Lowest price, lightweight
General Research	xArm 6	Balanced performance and price
VLA/AI Research	xArm 7	7DoF flexibility, Franka alternative
Industrial Applications	UFACTORY 850	Highest precision, extended reach

UFACTORY xArm Series

UFACTORY xArm Series

Overview

Company Introduction

Key Significance

Democratization of Research Accessibility

Comparison with Franka

Product Lineup

Model-Specific Features

Lite 6

xArm 5

xArm 6

xArm 7

UFACTORY 850

Common Features

Software and SDK

Official SDKs

Code Example

ROS 2 Support Features

VLA Research Applications

Open X-Embodiment Dataset

RT-X and VLA Models

LeRobot Framework

Research Use Cases

Accessories and Peripherals

Purchase Guide

Recommendations by Research Purpose

Where to Buy

References

Official Resources

Research Papers and Projects

Additional Resources

Footnotes

See Also