To all the makers out there -

This is a work in progress. We will update it with a complete "How to Make" description as we proceed. If you want to build it with us, please let us know in the comments section.


Sanju Mathew

Evolve is a plug and play platform for Prosthetics and Human Machine Interfaces. Currently developed for a 3D printed Bionic Arm, evolve features an App store so that users can download new features and install it into the arm and a a range of specially designed sensors to know the exact pressure required to hold an object - whether its an apple or a glass bottle and Neural Network for AI based learning.

The current version of evolve was tested to work as a Human Machine Interface, on an inhouse custom built Bionic Hand and on the Inmoov hand (www.inmoov.fr).

The following instructable is written based on how we interfaced EVOLVE to a Bionic Hand

The control input for the bionic hand is a set of EMG sensors and works both as a wired and wireless input. The GUI of the hand gives basic insight into the status of the hand, the apps that are running and features installed. A profile based system and touch input allows the user to select the required functionality for his hand, whether its a key grip to hold a car key or a power grip to hold a ball.

Until now, all the 3D design, hardware development and software coding was done by the same team. We hope to bring a change in this by providing a hardware platform, open source 3D hand model and and app ecosystem. With this, developers working on the 3D model can completely work on it, apps and features can be developed by software developers and hardware teams can develop accessories for the system. This helps to bring down the cost of the system and make it scale able, smarter and easier to work with.

We started working on a bionic hand in 2010, when we first came across a 30,000$ bionic hand video and realized that the number of amputees in need of an affordable version is really high.

Step 1: Power Supply

The motors being used is a combination of MG996R and Each of the motors require around 1 Amp of current and runs on 5V supply. So the basic design consideration is to have a 5 Volt 6 Amp supply powered from a 3.7 Volt 7000 mAH Li-Po battery.

To achieve this, we used a simple boost converter based on XL6009. The demo was done using a 9 Volt dc power adapter and a buck converter.

Step 2: The Sensors

Single Supply EMG sensors were used as the control input. Currently the system is built with specifications that is suitable for both gelled electrodes and reusable electrodes. Its a wireless wearable device that monitors the users muscle signals and transmits a DC averaged value back to Edison via a BLE interface. It also has an option to transmit a partially processed EMG signal after 16 bit ADC conversion (ADS1115).

Step 3: The Brain

When the Intel Edison came to the market, most of our hardware requirements in terms of wireless communication and a powerful processor and memory were immediately satisfied.

The code was done in python script and GUI was built in Node.js. Intel edison sits inside the bionic arm and still gives us the power to have a wireless device interfaced with the hand running python and node.js. With python we were able to develop a system to where we can download new apps and features and gives the user an ability to use/run it as required. This entire system was built from scratch.

Step 4: The Motors

A combination of MG 995 R and 996 R were used. Head on to the HobbyKing website to get more info. about this.

Step 5: The Arm

The acrylic version shown above is our first mock up model.

A detailed assembly instruction of the Inmoov hand can be found in the project website - www.inmoov.fr

Our in house 3D design will be updated shortly.

My son lost his hand at age 16. We have been buying/replacing $15,000 hands for 5 years now. This is definitely my next project!! Do you have more info about controlling the movements?
<p>Sorry to hear about what happened to your son. The huge prize of bionics out there is one of the reason why I started working on it 4 years back. Let me know if I can be of any help to you.</p><p>At present controlling is done via EMG sensors attached to the remaining muscles of the person. We have built a 6 channel EMG sensor for Evolve. Depending on how much muscle is remaining and what number of them can be actuated, you can choose the number of EMG sensors to be used.</p><p>If it is just a single EMG sensor, a single flex of the muscle will cause the hand to close and double flex to open. A single flex in between will pause the action (close or open) being performed.</p><p>Evolve also has a feature to download and install more 'Features' like key grip power grip, etc. This can be selected via a touch sensor at the palm of the hand.</p>
<p>I love to see this kind of work being done, thanks for sharing this!</p>
<p>Thank you for those encouraging words :)</p>
Do you have a website with the full details? Software, sensors, design details etc?
<p>Not as of now. We built the current version during Intel Hackathon and is working towards completing the hardware. Will post it once its done. If you are trying to build one, I can try help you out and send you the details as and when we have it.</p>

About This Instructable




More by StageRnD:Evolve - A platform for Bionic Hands and HMIs (Intel IoT) 
Add instructable to: