NVIDIA’s High School Interns Spend Summer Creating Robots

by Ahana Dave

High school’s usually closed during the summer, but our inaugural batch of young “Jetson” interns never stopped learning.

Over the course of seven weeks, we hosted 12 high schoolers to work with our embedded computing group, which focuses on intelligent machines, like drones, robots and submersibles.

Age and location weren’t an obstacle. The youngest, Isaac Wilcove, 15,  just finished his freshman year and has been programming since he was eight. Others, like Michael Chacko, who hails from Texas, moved to Silicon Valley for the hands-on experience.

The interns were divided into three teams, each tasked with building Jetson-powered robots. The robots can perform various functions for business facilities, and may one day be used in a major new building nearing completion at our Silicon Valley campus.

But it wasn’t always smooth sailing: the young creators grappled with obstacles involving design, code and motion.

Team Electron

Grace Lam, Nicholas Liu, Mokshith Voodarla and Ravi Krishna from Team Electron developed an autonomous delivery robot that transported food, drinks and other items between cubicles.They deployed simultaneous localization and mapping mechanisms using Turtlebots and taught the machine to navigate around the office.

The robot initially had trouble determining its position within the office because the floorplan fed into the robot didn’t encompass hurdles such as trash cans. That’s when the Jetson AI module proved to be a smart assistant.

“With the Jetson, we have more computing power to run neural networks and visualization tools that allows moving and mapping at a higher speed and localize better,” said Lam, a junior at Palo Alto High School.

jetson intern autonomous car

Team Trashformers

With a focus on picking up discarded cups around the office, Team Trashformers programmed a robot to collect garbage and deposit it into the nearest trash can. Chacko, Shruti Jajanathan, Ishan Mitra and Mark Theis trained a Robotis Bioloid GP to recognize the refuse using photos of white styrofoam cups, positioned at various angles.

One of the main obstacles the team encountered was balance. The robot needed a Jetson, a battery and a camera so the team could coordinate its directions. But that made the robot front heavy, said Jajanathan, a junior at Cupertino High School, so the team came up with a workaround.

“We’ve had to compromise our motion,” said Mitra, a senior at Monta Vista High School. “Instead of turning, the robot is shuffling.”

Another challenge was the robot’s physical structure. The “arm doesn’t have range that a human arm does,” according to Theis, a senior at Woodside Priory School, so it focuses on picking up cups sideways.

Team Formula 1 Epoch

Team Formula 1 Epoch, with Abhinav Ayalur, Ricky Avina, Lynn Dang and Wilcove, trained an MIT RACECAR to perform the role of a fire marshal. In case of an emergency, the JetsonHacks car will go around a building and account for people, while reporting its results online.

The car’s neural networks required a lot of data to function. Data is transferred between a controller and a camera, a link that had to be rewritten for better connection by Dang and Wilcove, from Evergreen Valley High School and Mountain View High School, respectively. The team also ended up switching to a web camera instead of using two different cameras because of software bugs, Dang said.

“I think one of the big things is that even though we’re kids, we’re proving to the world that we can do something,” Ayalur said. “We can do what seemingly before only adults, Ph.D. people, can do.”

Friday Funday

It wasn’t all work for the group. They had lunch with NVIDIA founders Jensen Huang and Chris Malachowsky, attended speaker series, and stuffed themselves with donuts every Friday.

“This environment at NVIDIA is really cool with technology that’s leading the way in AI and all the engineers who are so willing to help us with their background,” Liu said. “It’s encouraging to see what the future is and what we’re going to be stepping into.”

Murali Gopalakrishna, our head of product management for intelligent machines, recruited and worked closely with the interns. He said he “realized why I became an engineer after seeing these students” and was impressed with the level of understanding and independence demonstrated by the interns.

All the interns are part of robotics teams outside of NVIDIA, which helped them bond and work well with each other, said Voodarla, a senior from Cupertino High School. Between breaks, they often held robot races.

The team’s code has been shared on GitHub, an open source website, for students and professionals looking to use Jetsons on their robotics projects.

(Writer Ahana Dave interned on NVIDIA’s corporate communications team in the summer of 2017.)