Yes, I am! And it’s not as overwhelming and scary as you think it might be, if you just approach the problem right. My two research students for the summer, Kenner and Brooke, are working on developing a prosthetic hand that can be controlled by signals from the brain. Seriously, they really are!
The key is to break the overwhelming problem down into manageable pieces—truly the key to any daunting task you might face. The two main pieces to our problem are 1) the prosthetic hand and 2) the control algorithm from neural signals. Each of these can then also be broken down into even smaller pieces, until we have manageable, weekly goals to accomplish.
Prosthetic hand—designing and building a prosthetic hand that works and looks like a human hand is no small feat. Several researchers have been working on this for years at no small expense. And to be honest, there are others out there that are much closer to producing this than we perhaps ever will be. But our goal is not to create a prosthetic hand that we can sell to the highest bidder, but to create a simplified device that can be controlled with a simple algorithm derived from the very manner in which our brains control our hands. Essentially, we would like to reproduce the 14 degree of freedom (DOF) data that has been previously collected in primates (rather than the 20+ that are actually in the hand) using a robot. However, even producing a 14 DOF robot that sort of looks like a hand is a difficult task. Kenner is currently working on a virtual solid model of each of the segments of our hand and the overall assembly of those components in the computer so we can print them with our rapid prototyping machine. He is also investigating and planning the components (motors, wires, microcontrollers, etc.) that we will need to develop the electronics of the robotic hand. This hand will likely go through several iterations in the coming years.
Control algorithm—developing an algorithm to control our simple robot is perhaps the most complex and wide-open part of our project that will require many different steps along the way. Researchers have been going about similar problems (for example, controlling a robotic arm using neural signals) for several years, and have done it several different ways. Our goal is to try to create as simple an algorithm as possible that mimics how the brain controls the hand, and that utilizes as few input signals as possible. My previous work has found that patterns described by principal component analysis are a possible route for doing this. Currently, Brooke is working on refining my principal component work, regressing the principal components with neural signals, and comparing her findings with my previous findings so that we can see if it is an improvement or not. The next step will then be to essentially do the regressions in reverse, or use the neural information to PREDICT the hand shape. Very exciting stuff!
So, yes, slowly but surely, we will do this!