This story was originally published by U of T Engineering News
By Samantha Younan
U of T Engineering’s self-driving car team aUToronto has once again taken the top spot at a prestigious international challenge — and notched a few personal bests as well.
The AutoDrive Challenge II invites ten teams from universities across North America to compete at the Mcity Test Facility in Ann Arbor, Mich. each June.
Though this year’s competition involved several new challenges and levels of difficulty, the team maintained their track record of success. As a bonus, a paper about the custom algorithm developed by the team members was accepted into Canada’s top conference on AI-robotics innovation.
“We’re always proud of the team, but to see them improve so much from year-to-year and for their hard work to be accepted for a conference publication really shows how beneficial club work can be for students’ professional goals,” says Professor Tim Barfoot (UTIAS), one of the academic advisors to the aUToronto team.
Each year, in late August, the competition releases a new set of rules and challenges. The team then has roughly 8 months to get to work before heading to the competition. Even for a team as experienced as aUToronto, each year comes with new surprises.
“There was a big jump in difficulty. I feel like this happens every other year and this year that jump was quite high compared to what had I experienced before,” says team captain Chad Paik, a PhD student at UTIAS.
One of this year’s biggest challenges also led to the team’s biggest success.
“With the Buy off Ride Challenge, essentially what happens is you start the car somewhere and then you have to hit six destination points out of a possible 30. The car has to go from one [destination] to the other, but the key here is that the challenge is almost 20 minutes long,” says team member Connor Wilson (Year 4 EngSci).
“Your car has to stay autonomous for 20 full minutes without hitting anything or without you having to take over. That raises the bar a lot, because previously they were five-to-ten-minute challenges.”
Not only did aUToronto receive the top score in this contest, but they were 150 points above second place.
All of the challenges are carried out with relatively low computing power in the car compared to some state-of-the-art systems — something the team says is both a hindrance and an advantage.
“I think that forces us to innovate in different ways,” says Wilson.
“For example, the [the planning and controls portion] that I worked on, there’s no AI involved, but I was still able to kind of find some novelty there.”
With pressure to remain on top after several years of wins, the team employed some new strategies.
“We had one milestone presentation [with faculty advisors] every two months. The desire for the students to showcase their abilities to the professors really help with keeping everybody in check,” says Paik.
The team was also committed to testing as early and as often as possible, with each of the different events they took part in giving them opportunities to demonstrate what they were working on.
While these demos were helpful, nothing prepared the team for what would happen at the competition.
“On the day of the competition, a hardware failure we had never experienced before happened… the computer just turned off. For the next challenge, the GPS unit also failed,” says Paik.
“We were devastated.”
Despite the setbacks, the team was able to fix both the computer and GPS at the last minute and move their run to the end of the day.
“To see them under that much stress, and not only support one another and pull through but do so well on that challenge. It’s an inspiring moment to witness and I hope they’re very proud,” says Professor Steven Waslander (UTIAS), academic advisor to the aUToronto team.
Wilson, Paik and others have published a paper that outlines the custom algorithms they developed for the planning and controls portion of the competition. The paper was presented at the 2025 Conference on Robotics and Vision, in Calgary, last spring.
Their work focuses on autonomous driving in busy urban environments by unifying path planning and vehicle control into a single framework. This hybrid approach ensures that route generation and execution are optimized for both safety and navigation.
This is the team’s seventh first-place finish in eight years. Both Wilson and Paik will return next year and are already anticipating the 2026 challenges.
“Autodrive is something where I think you get out exactly what you put in,” says Wilson.
“There’s a lot of opportunity to learn. The more time you’re spending on it, the more you’ll learn.”