OG for the category of dexterous hands: SEED UK and its subsidiary product RH8D


About Seed Robotics

  • Seed Robotics is a British company founded in 2015, specializing in research and solutions for robotic dexterous hands, primarily in the fields of artificial intelligence, neuroscience, psychology, and cognitive science.
  • Their main products include: 3-finger dexterous hand RH2D, 4-finger dexterous hand RH7D, 5-finger humanoid 1:1 dexterous hand SR-RH8D, and FTS tactile pressure sensor series (which have released new possibilities by adding high-resolution tactile force sensing (1mN). These sensors can be embedded in dexterous hands or used separately.

It starts with an idea…

Pedro Ramilo is the Chief Software Architect of the SEED team, who first came into contact with humanoid robots in 2007. At that time, humanoid robots such as NAO and Darwin OP2 were very popular, and their figures were basically visible in research institutes that studied robots.


So how about handing over your hands to the popular NAO robots and Darwin OP robots at that time? What opportunities can it unleash? Therefore, the first generation RH4D was born.


Years later, they launched a series of human inspired dexterous hands (adult and child sizes) and advanced dexterous hands with end effector tactile sensors. The following image shows the iCub robot (iCub is an international project on humanoid robots developed at the Italian Polytechnic University in Genoa, Italy. It is 108cm tall, weighs 25kg, has 54 motors and 76 degrees of freedom throughout its body)


At this point, the earliest commercially available humanoid robots in the world (NAO, Darwin OP2, and iCub) have all used the dexterous hands provided by SEED company.



Name:1:1 Five finger dexterous hand    Model:SR-RH8D

RH8D has 19 degrees of freedom, including the contralateral thumb and a complete spherical wrist joint. In addition, the three fingers are independently controlled by intelligent actuators. Underactuated design aims to achieve precise user control while also taking into account adaptive control for objects of different shapes.
The RH8D humanoid hand possesses advanced mechanical capabilities and real-time high-frequency data feedback, including position, speed, PWM output, and high-resolution current measurement. By combining these parameters, users can infer various information, such as estimating the magnitude of the force
Specially designed fingers can achieve magnetic detachment, using Dyneema (a Kevlar fiber) tendons and reinforcing the outer shell and elbow attachments, ensuring the safety and durability of daily use by users.

  • Dexterity 

Featuring 8 compact and powerful actuators, users can individually control:

1. Wrist rotation, wrist flexion, and wrist adduction;

2. Thumb adduction and thumb flexion;

3. Index finger flexion;

4. Middle finger flexion;

5. The joint action of the ring finger and little finger is balanced with tendon tension (each finger is consistent with the shape of the object).




  • Perceived

The advanced EROS control board and intelligent actuators provide real-time, high-frequency feedback up to 120 Hz (provided by

All data for driving joints can be obtained:

1. Location;

2. Speed;

3. Temperature;

4. PWM output (directional);

5. High resolution current measurement.

The TOF distance sensor on the palm can detect the proximity of objects. By combining available numbers

According to, users can infer other information, such as force estimation. Capacitive gaskets can be selectively installed

On the back of the hand, to gain richer H.R.I experience.


The above figure shows the data obtained by gently gripping the ball


The above figure shows the data obtained by forcefully gripping the ball

  • Controllability

Power supply voltage: 9V to 24V; The gripping force of each internal actuator can reach 6kg/cm, providing a payload (vertical tension) of 2.50kg; Payload (3D space): 750g; Flexible hand weight: 620g (all actuators are included in the unit).

When assembling on the robotic arm, if possible, rotate the DoF around the built-in wrist and use the arm's own rotation DoF.


Fingertip 3D tactile sensor

  • 1. High resolution -1mN resolution (0.1g/0.0035oz)

  • 2. Low cost

  • 3. Measurement range - up to 50N

  • 4. High overload tolerance and immunity to magnetic fields


  • Open source

1. Open source ROS software package;

2. Python (PyPot);

3. Low level serial protocol (UART).

4. Interface: UART, RS485 (or full duplex or half duplex TTL compatible with Robotis system);

5. USB interface: used for configuration and maintenance;

6. Optional Bluetooth: online maintenance and diagnosis.

7. Embedded processor: Each intelligent actuator of the 32-bit ARM Cortex M4 has a dedicated coprocessor.

8. Provide ISO 9409-1-50 installation adapters (suitable for most robotic arms), and custom adapters can also be produced according to customer requirements.


Legal Notice: Seed Robotics only provides 3D CAD files to its customers (entities that have purchased or ordered Seed Robot products) for the purpose of constructing simulations and evaluating dimensions and mechanical characteristics.

Seed Robotics reserves all copyrights and intellectual property rights related to the distribution materials and makes no guarantees of their accuracy or applicability. The publication of this material does not constitute a waiver of Seed Robotics copyright and intellectual property rights, nor does it grant users any distribution, transfer, ownership or any other rights beyond those expressly authorized in this notice.

Seed Robotics:https://www.seedrobotics.com

More product information:https://kb.seedrobotics.com/doku.php?id=rh8d%3Ahome




Donghu Robot Laboratory, 2nd Floor, Baogu Innovation and Entrepreneurship Center,Wuhan City,Hubei Province,China

Technical Support

ROS Training

About Jingtian

About Us
Join Us
Contact Us

Cooperation and consultation

Business cooperation: 18062020215


Pre sales technical support:

Tel 13807184032

Website record number:鄂ICP备17004685号-1 | Technical Support | Contact Us | Terms of Service and Privacy | Map