Neuroprosthetics

Another one of my many projects from senior year of college, was centered around neuroprosthetics (brain controlled prosthetics), with a parallel project looking into artificial muscles. Above is my end of year video outlining my progress, and below is my final report. 

   

Executive Summary

This report outlines the beginning stages of this project, working towards a fully functional neuroprosthetic, or “brain controlled” prosthetic. By developing with an open source system, there is easy access to documentation and a high level of technological compatibility, which allows for faster innovation and implementation. This semester's work consisted of establishing a data stream from the OpenBCI hardware, and linking that to a servo motor for attention based actuation. Along with that, there was a portion of time spent developing and testing artificial muscles to increase fluidity and lower the weight of the prosthetic hand. The work done provides a solid foundation for developing a brain actuated prosthetic, powered by artificial muscles. 

Background

There are nearly 2 million people living with limb loss, in the United States alone [1]. Prosthetic limbs can range in cost from about $5,000 - $50,000 [2], before insurance. The majority of those prosthetic limbs are very low tech, and cause the user to adapt their behaviour to properly use their artificial limb. The ultimate goal is to have an artificial limb that can be controlled by the mind, and function just as well or better than a normal human limb. There have been advancements in this field, but the technology isn’t accessible to the consumer. There are some prosthetics that utilize muscle signals to move hands and feet, however the advantage of using a brain computer interface (BCI) is the modularity, and eventually the speed of signal transmission. If this were to be integrated into a system such as NeuroLink, the prosthetic would likely be able to move instantaneously with thought. That technology paired with artificial muscles could pave the way for prosthetic limbs that function at a higher level of strength and precision than organic limbs. 


Technical Work

The first task of the project was to establish a connection between the computer and the open BCI hardware. This proved to be quite difficult in the beginning, but after much trial and error, there was a steady stream of data as can be seen in Figure 1. 

Figure 1: OpenBCI Data stream 

After establishing a steady stream of brain data, the open bci hardware was interfaced with an arduino uno through the COM 4 and COM 5 ports (code and documentation in Appendix A). The particular widget in use was the binary “focus” widget, which essentially reads attention values of the brain, and sends a signal when the threshold is reached. This particular threshold was 25, and the range was from 1 to 100. Once that widget was working consistently, the signal was sent to the arduino to actuate a servo motor after the threshold was reached. It then was necessary to develop a testing setup, to verify that the motor would be able to move a prosthetic limb. For the purpose of this experiment, a small prosthetic finger was developed (Figure 2) then 3D printed, to be attached to the servo by a string. In order to make the finger move like a regular one, a length of elastic band was run through the backside to act as a ligament. As can be seen in Figure 3, the testing setup was fully functional, and allowed for free movement of the finger with the servo, as well as the artificial muscles.

Figure 2: Model of Prosthetic Finger

Figure 3: Testing setup

Along with inducing a motor movement with a brain signal, a parallel path was pursued in the region of artificial muscles. Artificial muscles allow for a higher strength to volume/weight ratio than motors, thus are an excellent option for prosthetic limbs. These particular artificial muscles are also significantly lower in cost, being made from regular nylon fishing line. The line is supercoiled with a fixed weight on one end, then tempered and stretched in the oven for two, 1 hour long cycles at 380 F. Once they cure, the muscles contract when exposed to temperatures above 70 C. 

Figure 4: Artificial muscle made from supercoiled nylon fishing line

As can be seen in Figure 5, the test setup also worked for the artificial muscles. The purpose of this project was solely to prove the viability of both options, thus the testing setup isn’t representative of a fully functional prosthetic. For this trial, a heat gun at 100C was used to contract the muscles. This proved to be effective, however the future of this technology will depend on an adequate thermal management system for heating and cooling the muscles. 

Figure 5: Test setup with artificial muscles contracting prosthetic finger

Future Work


The immediate future work of this project will be in developing a directory of brain signals, and a machine learning platform to recognize them to allow for localized control of a prosthetic hand (moving fingers, or the wrist). It would be advisable to onboard a computer scientist who is competent in signal processing and machine learning. Additionally, there is a need for a thermal management system that would allow for rapid heating and cooling of the artificial muscles. Figure 6 depicts a concept developed using a peltier module, and a custom heat pipe to rapidly transfer heat to and from the muscles. There would be multiple muscles coiled or braided around the pipe to allow for sufficient contact area, thus heat transfer. However, this is only one possible path to the final design. It is possible that a compressed air, or coolant heat exchanger could work. 

Figure 6: CAD of concept thermal management system

Conclusion

This project will increase the quality of life for those living with limb loss, by providing an open source, low cost, highly ergonomic, strong, precise prosthetic limb. The past semester proved that a motor could be actuated by brain signals, and that an artificial muscle could be used to contract a finger in a prosthetic limb. The future of this project will involve combining those two systems into one; a prosthetic hand, powered by artificial muscles, activated by brain signals.  Utilizing the OpenBCI platform to process brain signals will allow for quick prototyping and development of this system, and the artificial muscles will provide a new standard of strength, fluidity and precision to prosthetics.