Two people stand on a miniature track while a car navigates the course.

Those looking for a challenging but fun way to develop their critical thinking and advanced machine learning skills can take part in Humber’s annual AWS DeepRacer competition.

AWS DeepRacer is a fully autonomous racing league with a 1/18th scale race car driven by a machine learning (ML) algorithm. The students are tasked with programming the vehicle how to drive on a cloud-based 3D racing simulator. They can test the effectiveness of their programming, adjust as needed and then see the impact it has on their lap time.

When the competition is ready to happen, the model cars are taken from the virtual track to a real one where they race to determine which has the best time to declare a winner.

The inaugural 2023 AWS DeepRacer competition was a successful event that had students of all skill levels getting hands on with machine learning.

Mihai Albu is a professor with the Faculty of Applied Sciences & Technology (FAST) and a researcher at Humber. He was asked to help start and oversee Humber’s AWS DeepRacer competition and agreed, saying it’s a fun experience and something he wanted to be a part of.

Two people are holding a device while working on a small car that’s sitting on a table.

Last year, the competition was just for FAST students. This year’s competition was open to the entre Humber community. Currently, Humber has 10 scale cars that were donated by Amazon Web Services (AWS). The competition takes place in the spring with meetings starting in the winter for students who wish to participate.

Albu, who teaches the Artificial Intelligence with Machine Learning program, said taking part in the competition is a valuable experience for learners. They can develop their programming skills while gaining practical experience that can be included on their resumes.

It’s also teaching the students about the concept of reinforcement learning, which is a ML technique that trains software to make decisions to achieve the most optimal results.

Albu said the students learn as they experiment with different parameters in the programming and can see the results of their decisions in real time based on how their model does on the lap.

“They see the results and they want to try making more and more changes to see if that improves their time,” said Albu. “It’s amazing to see their reactions and the passion they have for this. They’re often like ‘Wow, I managed to get a better time so why not try more changes and see what happens.’”

While it’s still being decided, the hope is to expand next year’s event beyond Humber to include other post-secondary institutions.

Albu said AWS is a strong supporter of the competition and usually sends multiple personnel to the event as well as the meetings. Albu said they’re always willing to lend a hand or provide advice to the students as they navigate the challenge.

This AWS DeepRacer competition is open to students from all faculties and programs across Humber. Interested students are invited to send an email to Albu at mihai.albu@humber.ca expressing their interest in participating in the competition.