High School Club Builds Self-Driving Vehicle

Sophomore Ravi Krishna discusses how the project teaches students about machine- and deep-learning technologies

19 October 2017

“We are building a 1:10 scale model of a self-driving car that will be able to navigate autonomously around campus,” high school student Ravi Krishna wrote to The Institute earlier this year. “We currently have about 30 students working on it. Our goal is to learn about and experiment with machine-learning and deep-learning technologies that are so important to our future. As far as I know, ours is the first such effort at a high school.”

We frequently receive unsolicited pitches asking for coverage of all manner of technologies, so when we received Ravi’s message asking us to write about his self-driving car project, the ShadowCart [above], we were intrigued.

Ravi’s father is IEEE Member Suresh Krishna, so that’s how the sophomore at Los Altos High School, in California, is familiar with The Institute. Ravi said he hoped an Institute article about the self-driving car would inspire many of the IEEE members who are machine-learning and robotics experts to help the Electric Dreams school club he launched to work on the project.

The Institute asked Ravi to tell us more.

Why did you start the club?

Living in Silicon Valley, you see innovation happening everywhere. I developed a great interest in machine learning and self-driving vehicles, so I thought it would be really cool to share that interest with my peers. My classmate Maxwell Liu and I started Electric Dreams in February because there was no schoolwide effort to teach students how to use machine-learning and deep-learning technologies. Our club is a student-led effort. Its primary mission is to make machine-learning technology, particularly as it applies to self-driving vehicles, accessible to high schoolers. Our classmate Alexander Ng joined the club later as vice president. We split up duties including building the car, writing code, developing the computer infrastructure for deep learning, and building a website. Getting your hands dirty and actually working with the technology is the best way to learn it.

What hardware and software did you use to build the vehicle?

To make the project as accessible to students as possible, we used open-source code and off-the-shelf parts. All the electronics are mounted on a platform of a Traxxas radio-controlled car, which interfaces with the power-steering servo and motor. The components used include 2D Lidar to detect obstacles, an Nvidia Jetson TX1 processor to control the car, an inertial measurement unit, and a speed controller with odometry. It also has a Logitech camera to enable the car to recognize obstacles, USB speakers, and batteries. In terms of software, the car runs on Ubuntu Linux, and most of the software is written in Python.

How can this project benefit other students?

If students learn how to use these technologies, that’s a way to give them a huge advantage in the job market. To make things easier for a large group of students with varied skill sets to contribute to the project, we are building a more standardized curriculum for all club members to make sure that everyone is on the same page. This curriculum is split into several modules: servomotor control, image classification, object detection, path planning, and Lidar software interface.

Who taught you about the technology you’re using?

For the most part, I’m self-taught. I took a series of C programming language classes a few years ago from Stanford’s GiftedandTalented.com program. I learned the Python programming language on my own. In eighth grade, I took advanced-placement calculus and statistics courses, which enabled me to take multivariate calculus in ninth grade. We are fortunate at my school because we have a professor from Foothill College who teaches this course. This class turned out to be useful for understanding machine-learning algorithms.

I’ve finished Udacity’s online machine-learning engineer nanodegree. I also interned at Nvidia, in Santa Clara.

Besides the tech skills, what else have you learned?

I’ve learned how to get people to see a common vision and inspire them to work toward that goal.

Do you plan to pursue an engineering degree?

Yes. I would love to go to the University of California, Berkeley. It has an amazing computer science program. My dream job would be to start a company one day.

Tell us about your advisors.

They are Adam Randall, who teaches advanced-placement physics at the school, and Douglas Lublin, the professor of multivariate calculus at Foothill College. Also Barrett Williams, a former technical marketing manager at Nvidia, Kelvin Lwin from the company’s Deep Learning Institute, and Jeffrey Smith from Nvidia’s autonomous drone group. And Kurt Keutzer, a professor from University of California, Berkeley, whose group is doing leading-edge research in deep learning and autonomous vehicles.

What kind of support have you received?

We got a lot of help from Jeff Harding, the superintendent for the Mountain View–Los Altos (MVLA) school district. He stopped by one day, and I explained what we were doing and how this could really benefit the school. He took a huge interest in the self-driving car project—which meant quite a lot to me. He was instrumental in helping us obtain an innovation grant from the MVLA Foundation, without which the club would not exist. He has been supportive of efforts and even featured us at a district board meeting last year.

To learn more, read an article about the club in the Los Altos Town Crier.

IEEE membership offers a wide range of benefits and opportunities for those who share a common interest in technology. If you are not already a member, consider joining IEEE and becoming part of a worldwide network of more than 400,000 students and professionals.

Learn More