A few months ago, we had Patrick Pilarski on our Embedded podcast. He’s a professor at the University of Alberta, in Edmonton, Canada, working in machine learning, robotics, and biomedical.
I find all three of these areas fascinating. Machine learning is such an amazing and growing field, I’m starting to believe in the singularity. Robots make me feel like a five-year-old, giggling over their antics. And biomedical, well, I’ve worked on some of that and it has the potential to change the world. But all together? How can that be?
Machine learning combines well with robotics: if you want to have a robot pick the ripest peaches from a tree, you need to give it a camera, some intelligence, and some training. What one robot learns from this, other robots can nearly instantaneous use. And once you have a robot that can pick peaches intelligently, then manufacturing things or doing surgery on humans is straightforward; it is all pretty much the same skills with different applications.
But what about the biomedical part? Well, Patrick applies machine learning and robotics to prosthetic limbs. My jaw drops when I think of artificial limbs that know if it is 7am and you are moving around, you are probably getting coffee so it is time to move the prosthetic hand to grip a mug. That’s just the start of what he’s working on.
Whenever I think about Patrick’s lab, I am filled with feeling of glee, hope and envy for the information and technology. I want the world he’s building.
Tempting as it is, I’m not ready to move to Canada to pursue a PhD. So when I ended up with some free time, instead of filling it with work, I made the decision to spend a few weeks learning more about machine learning and robotics.
I wrote about building a robotic arm, learning a little about the amazing machine learning platform I was gifted, and learning to make it move on command. I want to make a typing robot because it would be hilarious and awesome. But also because once I can get two robot arms to type based on camera input, I think maybe I can start thinking about larger robots picking peaches (or other things).
To properly get started though, I need to name the system. I asked on Twitter and nearly named the project Sam Serif. That is about the right level of hilarity. But then @BentTronics sent me a post about Thomas Peterffy.
To summarize, Thomas wanted to use computers on the stock market but when he wired in a computer, people fussed. (This was years ago, before it became all computerized.) Upon being told that trades must be entered via keyboard, he made a typing robot, connected to his computer, that pounded on keyboard at intense speed.
After some discussion about naming the robot after Thomas Peterffy, its main purpose in life (to type), and the magic of palindromes, my robot’s name is Ty Peterffy Epyt. That’s a long name to keep typing so, of course, the middle name gets shortened to an initial: Ty P. Epyt (you can find it on github as TyPEypt). And here, on this blog, where we are among friends, it is just Ty.
I have made more progress on Ty this week. I learned about Robot Operating System (ROS) which is powerful and huge. Of course, with great power comes immense documentation. I’m working on that and hope to write up some basics in a post soon.