Watch: Human pilot, AI race their drones in Abu Dhabi; who won?

Khaleej Times

17-04-2025

A drone piloted by AI has convincingly beaten a human-controlled machine in an international drone racing competition in Abu Dhabi, marking a significant milestone in the development of artificial intelligence and autonomous flight.
It also marked a global first, where AI outpaced human pilots 'in a race of such scale, speed and complexity featuring some of the top drone pilots in the world,' organisers of the inaugural A2RL (Abu Dhabi Autonomous Racing League) x DCL (Drone Champions League) Autonomous Drone Championship said on Wednesday.
The AI-piloted drone of Team MavLab from Delft University of Technology (TU Delft), The Netherlands) outdone a world-leading human pilot to win the AI vs Human Challenge, one of the four race formats.
Team Mavlab also dominated two other races, including the AI Grand Challenge, where it set the fastest time on the 170-meter course by completing two laps (22 gates) in just 17 seconds.
The same team from TU Delft also claimed top spot in the Autonomous Drag Race, touted as the world's first AI-only drag race. Team Mavlab demonstrated straight-line speed and control under high acceleration against other top teams.
Meanwhile, TII Racing (Technology Innovation Institute, Abu Dhabi) won the AI Multi-Autonomous Drone Race, a high-speed test of AI coordination and collision avoidance.
The goal of the competition was to push the frontier of AI. The drone had access to just one forward-looking camera, a major difference from previous autonomous drone races. This is more similar to how human first-person view (FPV) pilots fly, and leads to additional perception challenges for the AI.
Head-to-head duel
The head-to-head duel between AI and humans was the most complex ever staged. 'With no human input, the drones relied entirely on real-time processing and AI-driven decision-making to reach speeds exceeding 150 km/h through a complex race environment,' the organisers noted.
Watch the video below:
The course design pushed the boundaries of perception-based autonomy— featuring wide gate spacing, irregular lighting, and minimal visual markers.
Each team raced a standardised drone equipped with the compact yet powerful Nvidia Jetson Orin NX computing module, a forward-facing camera, and an inertial measurement unit (IMU) for onboard perception and control.
The use of rolling shutter cameras (a type of image capture in cameras that records the frame line by line instead of capturing the entire frame all at once) further heightened the difficulty, testing each team's ability to deliver fast, stable performance under demanding conditions.
'This marked the first time an autonomous drone race of this scale and complexity was staged on such a visually sparse track, underscoring the ambition and technical challenge of the event,' the organisers added.
Christophe De Wagter, team principal of MavLab, said: 'Winning the AI Grand Challenge and the AI vs Human race is a huge milestone for our team. It validates years of research and experimentation in autonomous flight. To see our algorithms outperform in such a high-pressure environment and take home the largest share of the prize pool, is incredibly rewarding."
For two days, 14 international teams qualified for the finals week, with the top four advancing to compete across multiple challenging race formats. Teams from the UAE, Netherlands, Austria, South Korea, the Czech Republic, Mexico, Turkey, China, Spain, Canada and the USA represented a mix of university labs, research institutes, and startup innovators, and battled it out for the $1million prize pool across four race formats.
How AI won?
The team of scientists and students from TU Delft won the competition by developing an efficient and robust AI system, capable of split-second, high-performance control.
They noted: 'Whereas earlier breakthroughs, like AI defeating world champions at chess or Go, have taken place in virtual settings, this achievement happened in the real world. Two years ago, the Robotics and Perception Group at the University of Zürich was the first to beat human drone racing champions with an autonomous drone. However, that impressive achievement occurred in a flight lab environment, where conditions, hardware, and the track were still controlled by the researchers – a very different situation from this world championship, where the hardware and track were fully designed and managed by the competition organisers.'
Team Mavlab created one of the core new elements of the drone's AI that did not require to send control commands to a traditional human controller, but directly to the motors. The deep neural networks were able to mimick the outcomes of traditional algorithms with less processing time.
Real-life applications
'The highly efficient AI developed for robust perception and optimal control are not only vital to autonomous racing drones but will extend to other robots,' noted Wagter, explaining: 'Robot AI is limited by the required computational and energy resources. Autonomous drone racing is an ideal test case for developing and demonstrating highly-efficient, robust AI.'
Speed is a very important element since drones have a very limited battery capacity. This means, the faster they fly, the more distance they can cover.
'Flying drones faster will be important for many economic and societal applications, ranging from delivering blood samples and defibrillators in time to finding people in natural disaster scenarios. Moreover, we can use the developed methods to strive not for optimal time but for other criteria such as optimal energy or safety. This will have an impact on many other applications, from vacuum robots to self-driving cars,' Wagter added.
Meanwhile, the A2RL X DCL Drone STEM Program, designed in collaboration with Unicef and under the supervision of Advanced Technology Research Council (ATRC), has trained more than 100 Emirati students this year.
'At ATRC, we believe innovation must be proven in the real world, not just promised,' said Faisal Al Bannai, adviser to the UAE President for Strategic Research and Advanced Technology Affairs, and secretary-general of ATRC.
He underscored 'A2RL is more than a race, it's a global testbed for high-performance autonomy and reflects the UAE's commitment to advancing AI, robotics, and next-gen mobility responsibly.'