ROBOT NANO HAND

Once you have assembled your new handy desk top companion, program and run a hand gesture recognition demo, learn how to code and customize your own pose, all utilizing the inner palm camera.

The robot nano hand gives you the experience needed to understand the complexity of robotic design and the human form, create and build a humanoid robotic hand, and learn how to program and run a smart AI application.

Powered by the NVIDIA Jetson Nano Development kit, a powerful, easy-to-use, mini AI computer that lets you run multiple neural networks in parallel. It’s the perfect, portable solution for any project using image classification, object detection, segmentation, speech processing, and more.

The Robot Nano Hand is an open source project, designed and created by Rob Knight from therobotstudio, co-developed by Silicon Highway, and brought to life by Codethink.

 

The robot hand

The Robot Nano Hand, is an open source, 3d printed, robot hand with 4 independent fingers and a partially-opposable thumb, similar in size to an adult human.

Each digit is independently closed by a tendon and can also rotate from side to side at the knuckle. This results in the minimum number of servos for a strong grip with individual finger flexion and spread.

There are 11 DOF (Degrees of Freedom) in total. All the finger servos are housed in the body of the hand itself with an additional larger servo mounted below the wrist to provide pitch (back and forth) rotation of the whole hand.

Each digit is closed by a single tendon that runs the full length of the finger. This improves grip strength and stiffness by sharing the load over the full length of  the finger.

The tendon is contracted by simply winding it on to a 3d printed winch which is driven in turn by a small servo of the kind commonly used in radio controlled (RC) toys.

The specific model used (Feetech SCS009) differs from standard servos as it communicates via a serial bus rather than the legacy 50 Hz PWM (pulse width modulation) signal.

This allows additional operating modes and a greater range of motion better suited to driving tendons (300 vs the standard 170-180 degrees of rotation).

Additionally, these servos return data on their performance and temperature, which is important for the reliability of complex robots.

As there is only one powered DOF, the tendon, for the three joints of the finger, the tendon cannot fully control the position of the finger by itself.

In practice, a finger with just a  single tendon can comply to the shape of an object, as in wrapping around a bottle but it can apply very little grip force. Instead of evenly gripping the object, fingers like this tend instead to curl up to the smallest possible shape suddenly ejecting the object!
 

Print

There are two options to print the hand components, you can either print them youself by downloading the files from our Thingiverse page or for those without a 3D printer, the easiest, cheapest and quickest method to obtain the 3d prints is to order them online.

This also allows the use of SLS (Selective Laser Sintering) or other advanced printing technologies. Typically these produce higher quality prints and for one or two hands will work out cheaper than home printing.