Unitree releases new Dex5-1 humanoid robot hand

Touch sensors, tactile proprioception, design choices enable smooth operation

By Donald Halsing    May 28, 2025         

Unitree releases new Dex5-1 humanoid robot hand

Unitree Robotics

Unitree’s Dex5-1 humanoid robot hand is capable of touch sensing, proprioception, and can grasp up to eight pounds, with design features to help it grip and grasp objects.

Email Sign Up

Get news, papers, media and research delivered. Sign up for our free newsletters.

Stay up-to-date with news and resources you need to do your job. Research industry trends, compare companies and get weekly market intelligence with Robotics 24/7.

Robotics 24/7 newsletter
Unitree releases new Dex5-1 humanoid robot hand

Unitree Robotics

Unitree’s Dex5-1 humanoid robot hand is capable of touch sensing, proprioception, and can grasp up to eight pounds, with design features to help it grip and grasp objects.

Chinese humanoid and quadruped robot developer Unitree recently released its Dex5-1 dexterous robotic hand.

The company said its hand has novel advanced capabilities and is poised to open up new possibilities for what tasks robotic hands can achieve as end-of-arm tooling (EOAT).

Unitree also recently released a video of its G1 humanoid robots participating in a humanoid robot combat tournament.

20 DoF for smooth movement

The Dex5-1 features 20 degrees of freedom (DoF), of which 16 are active and four are passive. The thumb has four active DoF, while each finger has three active and one passive DoF

All joints support smooth backdrivability for direct force control, along with +/- 1 millimeter fingertip repeat position accuracy. Unitree said this hand model eliminates "stiff hands," making operations smoother and more convenient for reinforcement learning (RL) training.

Each hand is capable of +/- 22 degree four-finger lateral swing. Unitree said this ability can improve grip reliability and adapt more effectively to curved surfaces when gripping and grasping objects.

The rotation axis of each finger joint is close to the surface, reducing the gap between segments. Unitree said its “micro-gap” design enables a smoother and more fluid grasp, preventing joint corners from getting stuck on objects.

All five fingers can be replaced independently.

94 touch and temperature sensors

The Dex5-1P model adds 94 force sensors to the Dex5-1 base model. These include 10 in the palm, six in each fingertip, six along each finger, and six in the root of the four fingers except the thumb.

Touch sensors in the hand can perceive between 0.35 and 88.18 ounces (10 to 2500 grams). Pressure and temperature are measured.

Unitree said the sensors support secondary development of tactile AI algorithms for dexterous hands.

Power transmission supports tactile proprioception

The hand also features a high-power density hollow-cup motor and drive, high-precision encoder, and low damping small clearance reducer. 

Power is delivered through 12 self-developed micro force-controlled composite transmission joints that enable robots to achieve tactile proprioception (knowing where their limbs are in 3D space without looking) and four micro force-controlled joint gear transmissions.

The working voltage for the hand ranges from 24 to 60 volts, and it interfaces over USB 2.0.

It delivers perceptual and control feedback including joint mode, position, velocity, torque, temperature, stiffness and damping coefficients, along with voltage, current, and inertial measurements from the imu.

Grasp up to eight pounds

The Dex5-1 measures 8.56 inches from wrist to fingertips, 5.02 inches across its palm, and it 2.84 inches thick (217.3 x 127.5 x 72.1 millimeters). It weighs 35.3 ounces (1000 grams).

With the palm facing down, the hand can grasp up to 7.7 pounds (3.5 kilograms). With the palm sideways, the hand can grasp up to 8.8 pounds (4.5 kilograms). Its minimum grip diameter is 0.39 inches (10 millimeters).

The hand can tolerate up to 44 pounds (20 kilograms) of force and fingertips offer a strength of 10 Newtons. It can operate between -4 and 140 degrees Fahrenheit (-20 to 60 degrees Celsius).

Unitree Dex5 Dexterous Hand

About the Author
Donald Halsing, Associate Editor

Donald Halsing

Associate Editor

Donald Halsing is Associate editor of Robotics247.com. As an editor and journalist with a Bachelor of Arts in English from Framingham State University, he has a strong background in developing engaging and impactful stories for print and digital media. In addition to serving as Editor-in-Chief of Framingham State’s award-winning independent student newspaper, “The Gatepost,” Don spent over four years in operations at Mattress Firm, with his primary responsibilities including inventory control and inventory management. Don is currently pursuing his Master of Arts at FSU and is a professional photographer for Ashley Wall Photography

More about Donald Halsing

Latest in Reinforcement learning

Latest in Artificial Intelligence

Article Topics

Artificial Intelligence   Deep Learning   Machine Learning   Components   Grippers   Motion Control   Motors and Drives   News   Press Release   End-of-arm tooling   Force Sensing   Humanoid   Reinforcement learning   Unitree  

All topics

Editors' Picks

The future of CFD is connected, automated, and AI-enabled
The future of CFD is connected, automated, and AI-enabled

From geometry preparation to AI-assisted analysis, integrated CFD workflows…

Festo gets a grip on AI-based picking
Festo gets a grip on AI-based picking

Software-based GripperAI manages mixed picking through basic geometry

How Beckhoff Automation’s EtherCAT and controllers power Dexterity’s Mech ‘superhumanoid’ robot
How Beckhoff Automation’s EtherCAT and controllers power Dexterity’s Mech ‘superhumanoid’ robot

Safety, communication and motion control components enable smooth operation

Automate 2026: Forklifts, physical AI, vision systems and more from day three in Chicago
Automate 2026: Forklifts, physical AI, vision systems and more from day three in Chicago

North America’s largest robotics and automation event winds down