[MUSIC] I was a generalist as a kid. I loved all kinds of stuff. I thought I could be a scientist, I thought maybe a physicist, maybe a chemist, no idea really. I really was clueless as I grew up. I was a heavy athlete, I loved doing cross country skiing and mountain biking, these were the things I spent my time on. I didn't worry about the future. I had an idyllic childhood. And then you turn 16, or 17, or 18, and you start to think, I gotta do something fun with my life. And that's where I started picking up these challenges and realizing what it was that kept driving me. I loved solving math problems, I loved doing chemistry, or doing calculus challenges with my friends after school. This is the kind of stuff I did, just on the side, with no real focus. So it wasn't until I really started to try to pick something to do that I found these specific applications with the things I like doing, actually come up all the time. So as I graduated from university, I actually took a job at Pratt & Whitney Canada, which is one of our small jet engine manufacturing companies. While I worked at Pratt & Whitney, it was fascinating, we got to do things like test out jet engine controls. And so I had all this fun exposure to aeronautics, but the challenging part about aeronautics is that there's an extremely strong safety requirement. And so everything you do has to be buttoned down, and has to work precisely every time you do it. And so the development in that area really is dictated by 30 year old technology that's been vetted over, and over, and over. And so that was really frustrating for me, as a young engineer trying to get things moving and trying to change the world. So I took those wonderful experiences from the aerospace industry, and that focus on safety, and I turned it into a graduate career, where I entered the aeronautics and astronautics, and started building drones. Yeah, it's problem solving, it’s algorithm development. It's taking a robot task, something like landing a drone on a moving car, or trying to get a self driving car to park itself or navigate through an obstacle field. All of these problems are really practical, really applied, and you can really see the results of them as you're building, you can see how well it's working. If the drone misses the car, it crashes and fails, there is sort of an exciting aspect to that that's really fun, right? So to make a self-driving car, you have to first start by turning a regular automobile into a drive-by-wire vehicle. This means that you can plug a computer into the car and send commands to the gas, steering, and brakes, and the car will actually execute those maneuvers. Once you've got that capability, you can literally start to program it to do whatever you want. From there, you have to add a whole suite of sensors, and the computation in the trunk, and with all these pieces you can start to built up your own self-driving vehicle. So my research lab is built around the idea that if we have a real test bed, and we put it in a real situations, we'll learn a lot more quickly what the limitations are in current methods. What we're looking at, really, is finding research problems that are particularly suited to the academic community. So we're looking at those problems like dealing with adverse weather conditions, or dealing with some of these edge cases, or trying to theoretically demonstrate that the methods we're using have a safety aspect that we can quantify. So there's really lots of specific problems that we can work on in academia, we can try a bunch of crazy ideas. We can explore this space of possible solutions a little bit faster than, or maybe a little less guided, than you might do in industry. Definitely going to see increased performance from these vehicles. We're going to see things like the Tesla autopilot just getting better and better every year. And we're going to see fleets of vehicles that really are autonomous, working in restricted domains where they know they're safe to operate. But I still think that we have a lot of work to do to get that final full autonomy anywhere you want to drive it level. This is just because there's so many different ways that the cars can fail. There's so many different situations that they have to understand. So many strange conditions that you just can't predict ahead of time. So, if you're new to self driving, and you think this is a wonderful area, and you want to try it, I think the first thing you want to do, obviously, is follow this Coursera specialization. But from there, there's hundreds of ways that you can enrich your knowledge, and get a better sense of how to get into this area. This whole field depends on the fundamentals of robotics, the fundamentals of computer vision, the fundamentals of neural networks and deep learning. And these are areas that you can study extensively. And what's really fun about those is they also apply to other wonderful applications. So think robot manipulators, think cell phone camera systems, and improved features for taking videos, drone flights, and tracking of objects. There's tons of things you can do with this stuff. And for those of you thinking about whether or not self driving is a direction to go in and is a good avenue to explore, it's absolutely clear that this market is just going to keep growing. There's just an insatiable demand for the students that are coming out of our programs. It's pretty amazing, yeah. It's been a long time since I played with random math problems after school, right? And so to think that I'm now working on self driving cars, and I've been playing with drones for years, I'm just a lucky guy. [MUSIC]
