Skip Navigation

Students teach machines to race at OSU-OKC event

Students teach machines to race at OSU-OKC event

AWS DeepRacer event allows students to combine passion for robotics with machine learning

OKLAHOMA CITY (March 5, 2020) – Programming, machine learning and miniature-car racing came together at Oklahoma State University-Oklahoma City on Wednesday for one of the first events of its type in the state. 

Students from around Oklahoma competed to successfully program a miniature racer to drive around a track inside the university’s Community Impact Center as part of an AWS DeepRacer event.

AWS DeepRacer is an autonomous 1/18th-scale racecar designed to test reinforcement learning models by racing on a physical track. Using cameras to view the track and a reinforcement model to control throttle and steering, the car shows how a model trained in a simulated environment can be transferred to the real world.

Reinforcement learning is an advanced machine-learning technique. During the event, with the support of AWS engineers, participants learned about machine learning tools and applied what they learned to train their team’s robotic car to become self-driving.

Moore-Norman Technology Center student Tony Jordan said he was excited by the possibilities of machine learning in a programming competition.

“It’s a lot more interactive” than some other types of competitions, Jordan said. “You get to really learn what’s going on.”

His teammate Tyler Spicer fed an algorithm into the computer that he hoped would teach his racer to better stay on the track. Though he had worked with the Python programming language before, much of what he was doing was new to him.

“This is my first experience with machine learning,” he said.

Adam Carr, a computer science teacher at Bridge Creek High School, brought a combination of computer science students and robotics club members to the AWS event.

Marisa Jackson, who first got interested in robotics as a Girl Scout, has continued her interest into high school as part of the robotics team. This event seemed like a natural extension of that interest.

While she’s often on the build team for her robotics team, her troubleshooting expertise also translates to DeepRacer, she said.

Part of what she enjoys about robotics is “discovering the errors; finding out that certain things don’t work together.”

Carr said one difference between this and other robotics events his team has competed in before is, rather than give step-by-step instructions for the computer to follow, this event seeks to teach the machine.

“Here we are telling it what good behavior is, and then it teaches itself good behavior,” Carr said.

For example, good behavior in a racer includes staying on the track, and navigating the route as quickly as possible.

Bert Bishop, who brought his robotics team from Luther High School to the event, said activities like these help students see real-world results to their efforts.

“This gives them a chance to do something and translate it into real space,” he said.

 

###

 

For media inquiries:

Nick Trougakos, OSU-OKC – nick.trougakos@osuokc.edu – (405) 945-9196

About Oklahoma State University-Oklahoma City

Oklahoma State University-Oklahoma City (OSU-OKC) is part of the modern land-grant system of Oklahoma State University. OSU-OKC is a two-year public institution, offering more than 40 degrees and certificate options. Through quality education, a supportive environment, leadership training and service opportunities, OSU-OKC trains students to work and educates them to lead. OSU-OKC graduates are found throughout the community serving as police officers, firefighters, nurses, teachers, healthcare professionals and business owners. More information is available at www.osuokc.edu.