Racing with AI

The second season of the Abu Dhabi Autonomous Racing League (A2RL)is set to happen on 26 April 2025, at Yas Marina Circuit. Organized by ASPIRE, the innovation acceleration arm of Abu Dhabi’s Advanced Technology Research Council (ATRC),the second edition with an expanded pool of teams seeks to further stretch the boundaries of AI in motorsport. Khurram Hassan, Director of A2RL discusses how the competition helps test and advance autonomous vehicle technology

 Tell me a bit about the Advanced Technology Research Council of Abu Dhabi, the organization structure, and the vision behind this unique race concept.

The Advanced Technology Research Council (ATRC) based in Abu Dhabi is structured with a few key branches: TII, Aspire, and Venture One. TII focuses on world-class research, Aspire handles world-class competitions to drive innovation, and Venture One supports startups and R&D for commercialization. With Aspire, one of the competitions we lead is the Abu Dhabi Autonomous Racing League (A2RL). This league is centered on innovation in autonomous technology, aiming to push the boundaries of AI and robotics through high-speed racing.

How was the Abu Dhabi Autonomous Racing League was conceived?

The idea behind A2RL emerged as a concept to test and advance autonomous vehicle technology. Last year, we hosted our first race as a proof of concept, where eight teams from around the world competed in a fully autonomous, high-speed car race. The vehicles in this league are entirely self-driving and self-learning, pushing AI to its limits by reaching speeds close to 300 kilometers per hour. Much like how motorsport innovations have brought consumer advancements—think seat belts or braking systems—this autonomous racing league is designed to test and refine AI and autonomous technologies, preparing them for real-world applications.

With driverless taxis and robo-taxis emerging, is there a connection between those advancements and what A2RL is doing?

Absolutely. Technologies in robo-taxis rely heavily on AI to make real-time decisions in dynamic environments. Our autonomous race cars use similar technologies—LiDAR, radar, and cameras—but at a much more intense level. These vehicles are racing at top speeds, so they require precise, instant decision-making and advanced machine learning capabilities to navigate the track without any human control. If these systems can operate safely and effectively in such a challenging environment, we can confidently transfer these innovations to consumer and commercial vehicles, enhancing both safety and efficiency on the roads.

So this isn’t just an exhibition—it’s a testing ground for the technology, correct?

Exactly. If our AI systems can handle the conditions and speeds of a professional race, it shows the technology is mature enough for everyday applications. By refining algorithms in a controlled but extreme setting, we can help ensure that autonomous systems are ready for real-world use.

You mentioned something intriguing about a “driver angel.” What does that mean in this context?

The idea of a “driver angel” is about having an AI assistant available in crucial moments. While these cars are currently fully autonomous, it’s likely that, for a long time, human drivers will continue to rely on AI-assisted tools. This driver angel concept gives the car the ability to make the best decision possible in a critical moment, assisting rather than entirely replacing the human driver. It’s a step toward AI-driven safety that can intervene if it detects imminent risks or challenging scenarios, like sudden stops or an impending collision.

How does the cultural diversity among the teams influence the race?

That’s an insightful question. Each team brings its own unique cultural approach to safety and risk. Some teams prioritize time and finishing the race without incident, while others aim to outperform competitors at each turn. This diversity translates into different coding strategies, making each team’s AI algorithm distinct. We’ve attracted teams from places like the U.S. and Japan, and this global participation contributes to a richer development environment for autonomous mobility.

What do the teams actually receive to work with, and how do they engage with the vehicles?

Each team is provided with one of our Japanese Super Formula cars, which are the second fastest cars after Formula 1. These are equipped with an autonomous kit that includes sensors like LiDAR, radar, and cameras. We provide a foundational code and the vehicle, and the teams are responsible for customizing and refining this code to optimize their cars’ performance on the track. So, essentially, it’s a coding challenge, where the aim is to develop the best software to make the car navigate the track faster and smarter.

How have you seen the race and technology evolve from last year to this one?

Last year was all about proving the concept. We saw a lot of potential and realized there was a genuine demand for this as an AI sport. Technologically, we’ve improved the cars’ reliability and handling, especially for multi-agent, multi-car racing. For instance, in the first race, our autonomous cars clocked lap times of around three minutes, whereas human drivers were achieving times of about one minute and forty-four seconds. However, after three months of refining and racing, our AI managed to reduce its time to one minute and fifty-four seconds—shaving off more than a minute! So, it’s evolving rapidly.

Where do you host these races, and how has the public responded?

Our first race was held at Yas Marina Circuit in Abu Dhabi, where we attracted about 10,000 live spectators and over a million online viewers. We also had 22,000 people join virtually in VR, allowing them to experience the race from a 360-degree perspective as if they were the driver. The engagement and excitement were tremendous. We also collaborated with Meta to launch an app that allowed people to race against the autonomous cars in real-time, creating a uniquely immersive experience.

What’s the vision for the league moving forward?

We’re aiming to establish A2RL as the world’s first AI extreme sport, where humans spectate while AI-driven cars compete. This year, we’re expanding to 12 teams and building what we envision as a “mixed reality” sport. We’ll blend physical, on-track racing with virtual races that let fans participate digitally. There’s even a possibility for future collaborations with other major racing leagues like Formula 1, but for now, we’re focusing on the unique niche of algorithm-driven competition.

With all the innovation happening here, do you see AI taking over human drivers in professional racing?

Not immediately. Human skill and instinct are still at the heart of leagues like Formula 1. But as AI becomes more sophisticated, we may see it working alongside humans as an assistive tool. Last year, we held a race between a human driver, former F1 driver Daniil Kvyat, and an AI, and while Daniil won, the AI was only 10 seconds slower by race day. The car’s learning curve is accelerating, and who knows? AI might be able to match human performance sooner than we think.