Inside the Autonomous Racing League event that pitted a self-driving car against a Formula 1 driver
Explore the bustling pits of any premier motorsports spectacle, particularly the likes of Formula 1, and you’ll encounter a myriad of computer screens brimming with telemetry data. Today’s racing teams are immersed in a constant stream of real-time digital insights sourced directly from their vehicles. Having frequented these pits for years, I’ve always been captivated by the torrent of data displayed, yet never have I witnessed the Microsoft Visual Studio software development suite amidst the frenetic atmosphere.
However, my recent attendance at the inaugural Abu Dhabi Autonomous Racing League event unveiled an entirely different landscape. Known as A2RL, this groundbreaking event marked a departure from traditional racing formats. While the Roborace series focuses on solo car time trials and the Indy Autonomous Challenge revolves around oval track contests, A2RL embarked on a quest to pioneer new territories.
A2RL boldly introduced a revolutionary concept by fielding four cars simultaneously on the track, a feat previously unseen in autonomous racing. Moreover, the event took a monumental leap by orchestrating a showdown between the leading autonomous vehicle and a seasoned human driver—former Formula 1 ace Daniil Kvyat, renowned for his tenure with various teams from 2014 to 2020.
Image Credits: Autonomous Racing League
The true test unfolded behind the curtains, where teams assembled an impressively diverse ensemble of engineers. From aspiring coders to doctoral candidates to seasoned race engineers, each member grappled with pushing the boundaries in a novel domain.
In stark contrast to Formula 1’s realm, where ten manufacturers meticulously craft bespoke cars (sometimes leveraging AI), the A2RL race cars embrace a standardized approach to ensure equitable competition. Borrowed from the Japanese Super Formula Championship, these 550-horsepower beasts are uniformly designed, with no room for alteration of any component.
This uniformity extends to the sensor array, comprising seven cameras, four radar sensors, three lidar sensors, and GPS functionalities, all deployed to interpret the surrounding environment. Yet, amidst the hustle and bustle of the pits and conversations with various teams, it became apparent that not all are harnessing the full potential of the staggering 15 terabytes of data generated by each car every lap.
Take, for instance, the Indianapolis-based Code19 team, which embarked on the ambitious endeavor of crafting a self-driving car merely a few months ago. “There are four rookie teams here,” remarked Code19 co-founder Oliver Wells. “Others have been immersed in similar competitions for up to seven years.”
It’s all about the code
Image Credits: Tim Stevens
TUM, headquartered in Munich, and Polimove, based in Milan, boast extensive track records of success in both Roborace and the Indy Autonomous Challenge. Their wealth of experience, along with the transferability of their source code, has proven instrumental in their continued dominance.
Simon Hoffmann, team principal at TUM, emphasized the ongoing refinement of their code. While adjustments were made to enhance cornering behavior and overtaking aggressiveness to adapt to the unique demands of the road course, the underlying software framework remained consistent. “We continuously develop and enhance the code,” Hoffmann noted. “But essentially, we rely on the same foundational software.”
Throughout the weekend’s series of rigorous qualifying rounds, teams with substantial experience consistently topped the timing charts. TUM and Polimove emerged as the sole teams to achieve lap times under two minutes. In contrast, Code19 struggled to match their pace, with their fastest lap clocking in at just over three minutes, highlighting the steep learning curve for newer teams.
This competition presents a distinctive facet of software development seldom witnessed elsewhere. Unlike conventional coding challenges, where the focus may lie on debugging or algorithmic optimization, here, advancements in code directly translate into faster lap times and reduced accident risks.
Kenna Edwards, an assistant race engineer at Code19 and a student at Indiana University, exemplifies the transformative nature of this experience. Despite her background in app development, mastering C++ to develop the team’s antilock braking system posed a formidable challenge. Nevertheless, her efforts bore fruit, preventing potential crashes during the races.
The tangible impact of theoretical implementations extends beyond the racetrack, offering compelling career pathways. Edwards’ journey from interning at Chip Ganassi Racing and General Motors to securing a full-time position at GM Motorsports underscores the tangible value of hands-on experience gained through initiatives like Code19.
An eye toward the future
Image Credits: Tim Stevens
A pivotal aspect of A2RL’s mission is fostering development opportunities, particularly for younger generations. Running parallel to the main on-track action is a series of competitions designed for students and youth groups worldwide. Preceding the main A2RL event, these groups engaged in competitions involving autonomous 1:8-scale model cars.
Faisal Al Bannai, the secretary general of Abu Dhabi’s Advanced Technology Research Council (ATRC), envisions a progressive trajectory for these initiatives. He emphasizes a pathway where schools engage with smaller model cars, universities advance to autonomous go-karts, and eventually, teams transition to racing full-sized autonomous cars. This structured approach aims to inspire more individuals to pursue research and scientific endeavors.
The ATRC plays a pivotal role in supporting A2RL, covering expenses ranging from the cars to team accommodations. Their commitment extends beyond logistical support, as demonstrated by the extravagant festivities surrounding the main event, including concerts, drone races, and a breathtaking fireworks display.
Despite some technical hiccups, the on-track spectacle delivered moments of exhilaration. While the first attempt at a four-car autonomous race faced challenges, the subsequent race saw thrilling maneuvers, including a decisive pass by TUM, securing victory and the majority of the $2.25 million prize purse.
In the much-anticipated man vs. machine showdown, Daniil Kvyat showcased his prowess, overtaking the autonomous car twice to the delight of the enthusiastic crowd. The event attracted over 10,000 spectators, with an additional 600,000 viewers tuning in via live stream, underscoring its significance in automotive history.
Although technical challenges persisted, the event underscored the remarkable progress in autonomy. While the fastest car trailed Kvyat’s time by over 10 seconds, it demonstrated smooth, consistent performance—an impressive feat compared to the tumultuous beginnings of autonomous racing.
Looking ahead, the true measure of A2RL’s success lies in its ability to establish financial sustainability. While traditional motorsports rely heavily on advertising, A2RL offers an added dimension by fostering advancements in algorithms and technologies applicable to manufacturers’ vehicles.
Al Bannai highlights the ownership structure, where teams retain ownership of their code and can license it independently. This emphasis on algorithmic innovation positions A2RL as a potential catalyst for industry partnerships, showcasing the practical application of autonomous technology at speeds exceeding 160 mph.