The vehicle is tiny enough to traverse a hallway, maneuvering through a stream of ӽ紫ý computer scientists heading to their next class. Think RC car of childhood racing memories.
But this car has a brain of its own. There is no remote control-wielding human guiding it from point A to point B and around myriad obstacles along the way.
What it has instead: powerful algorithms.Those were created by an enterprising group of ӽ紫ý computer science students as part of a new and innovative ӽ紫ý engineering course that merges the physical world with cyber.
Completely driverless taxis recently took to the streets in San Francisco, Phoenix, and Las Vegas, carrying people to and fro despite concerns about safety — a sign that the future is here, but not quite fully developed.
That’s where Autonomous Racing: A Cyber-Physical Perspective (CS 489/689) comes in. The course, which debuted this fall, is a training ground for students who might want to be part of interdisciplinary research teams that will make driverless cars safer and more widespread.
“This course will not give them a complete education in how to build an autonomous vehicle,” said assistant professor Corey Tessler, who developed the course in collaboration with fellow ӽ紫ý computer scientist Prashant Modekurthy. “What it will teach them is that they need to fit into a larger team that builds [the car]. And that’s the way that all of technology always goes: more and more fields integrate with each other.
“This course is an excellent vehicle — pardon the pun — for getting that experience.”
ӽ紫ý graduate assistant Breanna Geller agrees. She started at ӽ紫ý as a pre-med major, but switched to computer science about two years ago. The 180 degree turn brought her here: to a course that’s pushing her outside of her comfort zone of computer science and into areas like robotics design and electrical engineering.
“This course is so unique,” she said. “I haven’t seen anything like this yet.”
Central to the course is an autonomous vehicle that’s 1/10th the scale of a regular car. Students will learn how to build complex algorithms that integrate with LiDAR and GPS sensors, and camera components that give the car the vision it needs to navigate the world, or, for now at least, ӽ紫ý’s campus.
Tessler and Geller envision the course as a launch pad for a student organization that will build cars to race in international . Similar to ӽ紫ý groups that have built energy-efficient homes, concrete canoes, and Baja racing vehicles, the new organization will give students a chance to apply their newly acquired academic knowledge to projects for national competitions. In addition to trophies, the students bring back resume-building experience in developing and testing future technologies.
And with F1 coming to Vegas this fall, their timing might just pay off.
“We’re at the edge of what you can do by going as fast as you can,” said Tessler. “This is an opportunity for students to really dive in and do research at the undergraduate level.”
Let’s take a spin, with Geller as our guide, around the new course that will have students racing to find the newest solutions for autonomous systems.
What is the ultimate goal of the course?
The outcomes are going to be a car that can race itself. We’re going to be split up into four groups, and throughout the semester, we’ll go head-to-head against each other over a series of three races to see which group is the better group, the faster group, the group that can put the best solutions on the car.
The ultimate test will come on Dec. 15 — our final race day.
We’re eventually going to try to get into international competitions. They’re called F1Tenth, and what sets you apart from other racers is the algorithms that you implement to the car. It’s not so much the physical attributes of the car. It’s all about the code.
How does the car work?
A key feature of our car is the LiDAR. As the car is driving, and at every so often nanoseconds, the 270 degree LiDAR is taking scans of everything that’s around it. The car uses that incoming data to determine its position, and, for example, how far away it is from the wall up ahead.
The car’s brain is NVIDIA Jetson NX, a tiny supercomputer module that runs the operating system that houses the algorithms that will be developed by students throughout the course. With our programs, the car can basically run itself.
How will the course help students to innovate new solutions?
There are different ways that you can tell the car, for example, to stop. So, right now we have implemented an emergency braking system on it so that when it’s running into something it’ll stop itself before it hits anything.
A different implementation that could maybe make the car go faster is focusing only on certain LiDAR scans, maybe the front of the vehicle, where we know we’re heading, and only stopping for those versus stopping for a parallel object that may not be in the vehicle’s way.
Since the course is so unique, is there a textbook to help guide students on their quest to win the final race of the semester?
No. is the No. 1 resource for students in the class. Much of the material is based on frameworks and materials authored by the University of Pennsylvania and shared with universities and schools through the open source platform. This course would not be possible without their contributions.
What are you most looking forward to about the course?
The different solutions that people come up with to solve problems that arise, especially with regards to competing and how to make the car go faster and how to avoid obstacles. The automotive industry is heavily invested in self-driving cars right now, and if you get started in building those algorithms and possible applications, you could come up with a really good solution, which could lead to a great opportunity after college.
And also, F1 is coming to Vegas! I hope it helps our course and our new student organization garner a lot more attention, because F1 is huge. I think deep down we all like to be competitive, we all like to race. So just having that opportunity to say, "This is my race. Maybe I’m not an F1 driver, but I can race a tiny little race car, an RC kind of car."
It’s so fun to drive the car. It makes all of the hard work so worth it.
What’s next?
I think we’re talking about setting up a follow-up course for the spring semester that focuses mostly on the racing component. You have the basics down pat, now, let’s race!
We’re also in the middle of setting up a registered student organization (RSO). We’re working on the constitution at the moment, and reaching out to other RSOs for guidance. We’re also looking for a catchy name. Rebel Racing was taken, so we’re open to ideas! [Rebel Racing is ӽ紫ý’s Baja car team.]
