Mind Controlled Robotic Prosthetics

Current prosthetics face several limitations that limit their accessibility and effectiveness. Traditional systems rely on rigid designs or myoelectric control systems that depend on nerve and muscle tissue data, creating barriers for many potential users.

Myoelectric prosthetics work by detecting electrical signals from remaining muscle tissue. However, many amputees cannot use these systems because they lack sufficient muscle tissue or have damaged nerves at the amputation site. This is common in cases of traumatic injuries, certain medical conditions, or when the amputation occurred high on the limb, leaving little muscle to generate detectable signals.

High costs, complex fitting processes, and limited availability in many regions further restrict access to advanced prosthetic technology. These challenges prevent many amputees from accessing the benefits of modern robotic prosthetics, particularly those who may not be eligible for current myoelectric systems.

This project develops a new generation of prosthetics that bypasses the need for muscle tissue entirely by utilizing EEG data for direct brain control. Instead of relying on damaged or missing muscle signals, the system reads brain activity directly, making it accessible to amputees who cannot use traditional myoelectric systems.

The solution implements advanced machine learning algorithms to interpret brain signals and translate them into prosthetic movements. This approach provides a cost-effective alternative to current systems while ensuring reliable identification of neural signals and focusing on safety and reliability.

The prosthetic arm was built using 3D printed components based on the InMoov open-source project. The design employs an innovative wire system that enables finger movements with minimal motors - each finger contains two wires (front and back), and pulling one wire creates grasping motion. The entire hand requires only 6 servos, significantly reducing weight and complexity.

The system uses an 8-channel OpenBCI Cyton board with dry electrode leads, positioned at FP1, FP2, C3, C4, P7, P8, O1, and O2. The board communicates wirelessly with the computer, ensuring maximum usability and minimal interference.

A RNN model processes the brainwaves to identify three distinct states: Open, Closed, or None. This third state helps reduce noise and improves accuracy. Commands are then sent to servo motors via Arduino for precise control of the prosthetic hand movements.

The BCI-controlled prosthetic system is designed for amputees currently ineligible for robotic prosthetics, particularly those facing financial barriers to current solutions. The system also serves users in regions with limited access to specialized care and those requiring simpler adaptation processes.

By utilizing EEG data instead of traditional myoelectric systems, this approach opens up prosthetic technology to a broader range of users who may not have the necessary nerve or muscle tissue for conventional systems.