AI trainees benefited from machine learning

Kenan Kao (left) watches Kevin Chor (center) and Atharva Joel (right) test AWS DeepRacers, autonomous vehicles used to teach developers reinforcement learning. They are three of about 24 summer STEM high school interns at GCU’s campus.

They hit and move and move and burn rubber, smoking other runners and leaving them in the dust.

But they were less of the glory of the compression, dash, move and burning rubber on the track than the glory of learning and problem-solving for their cars, two of the 10 Amazon Web Services DeepRacers that technology faculty recently purchased thanks to a $15,000 technology contribution from industry partner Discount Tire.

At least, that was kind of the glory of the Paradise Valley High School race car driver Atharfa Joel Arizona College High School Kevin Chor to imagine.

The Grand Canyon University The high school STEM summer internship students—two of about two dozen interns on campus studying everything from biomedical engineering to practicing science—spent the summer anticipating motor racing, which had recently begun to pay off in the Catalina classroom space adjacent to the Sunset Auditorium. and technology building.

GCU’s technology department used $15,000 from industry partner Discount Tire to purchase 10 AWS DeepRacers.

1/18 scale race cars teach reinforcement learning, a type of machine learning in which developers train machines, such as AWS racers, to make a series of decisions. In this case, Goel, Chor, and other AI/machine learning trainees are essentially teaching cars how to drive autonomously, then racing them on a cloud-based 3D racing simulator before moving on to racing on a real track.

The cars drive autonomously using cameras to display the track and the reinforcement model that students create and load into the car. According to AWS, the vehicle shows how a model trained in a simulated environment can be transferred to the real world.

“It (the technology) is exactly the basics of a Tesla car,” he said Jevon JacksonProgram Chair and Software Engineering Program Leader at College of Science, Engineering and Technology.

Students have spent the past few weeks using the Python programming language to “teach” cars how to drive themselves.

It’s just one of the projects published as part of the high school STEM summer internship program, run by K12 Educational Development in partnership with Honors College. In his second year at GCU, junior and senior high school seniors were integrated into the program with GCU professors and student advisors.

They were on campus five days a week and spent 100 to 200 hours during the internship, exploring STEM disciplines and attending professional development sessions organized by the Honors College.

Jevon Jackson

Jackson said the AI/machine learning interns, working with AWS DeepRacers, were learning something that was fun but also very challenging.

“It’s really, really complicated,” Jackson said of reinforcement learning. “Imagine professional athletes, people who, in their minds, see the bell bangs in their heads. It works in their mind, but then to put that into the actual game? This is a one in a million chance. You have to be extraordinarily good and physically practice this shot.”

“That’s exactly what machine learning is. It’s actually done in the brain – in a computer system. Then we take that and try to train hardware components (in cars) to work the way the brain says it is. It’s like a kid learning to learn and he has to do it more often than not. to correct himself.

And there was a bit of self-correction at the purple racetrack as Goen and Chor tried to solve why AWS DeepRacers wasn’t making turns on the track as it should. Meanwhile, a trainee in Artificial Intelligence/Machine Learning Stephanie Fana rookie at Desert Vista High School, has blasted around the track perfectly, at least twice.

It’s a scene that Jackson expects to repeat during the year as GCU students take turns programming autonomous prototype cars.

AWS DeepRacers is just a sneak peek at what awaits the college’s technology department. The process is beginning to add a focus on artificial intelligence to the software engineering master’s degree program, something Jackson expects to happen two years from now. The college also hopes to add an AI focus to its undergraduate degree program.

Head of Technology Programs Rob Lowe He said the technology department will also run the AWS DeepRacer competition for high schools. GCU students and faculty will instruct high school students to train their cars virtually, and in the spring, they’ll come to campus, load up their model in cars and physically race them on the track.

“I think today was the most frustrating And the The most fun day. This is part of how machine learning works. I think it’s really cool to see how reinforcement learning works and how these machines can learn themselves, just through trial and error.”

Atharva Joel, GCU Summer High School intern in STEM

The high school AWS DeepRacer program will be linked to the Spring International Christian STEM Competition, organized by GCU in partnership with the Association of International Christian Schools.

“We are learning a Many This summer, thanks to our high school interns, on how things do or don’t work — things we need to fix, as we set the track might not be the right kind of track,” Lowe said of the GCU-themed purple track.

He and Jackson suspect that they may need a darker track, perhaps black, to contrast the white fender lines and other road surface markings so that car cameras can better detect those markings.

Joel, who knew little about Python before he began his summer training, said that moving his model from the virtual to the physical one was the most challenging part of his internship thus far.

“I think today was the most frustrating And the He said watching DeepRacer crash into a wall multiple times because it didn’t turn around like it wanted it to. “That’s part of how machine learning works. I think it’s really cool to see how reinforcement learning works and how these machines can learn themselves, just through trial and error.”

Chor, who knew nothing of artificial intelligence and machine learning, said that self-driving cars are the future, and being able to delve into this technology makes it relevant.

“Loading into the actual car and seeing it move? It feels good,” he said.

Students only worked on their AWS DeepRacer models for about two weeks before putting their learning into practice.

“I’m so proud of them,” said Jackson, for seeing what they accomplished in that small window of time.

GCU Senior Writer Lana Sweeten-Shults can be reached at [email protected] or at 602-639-7901.