Formula 1 of heaven! Relive the incredible moment when an AI pilot BEATS three human champions in a drone race
Artificial intelligence (AI) in sports might sound like something out of a science fiction movie.
But incredible footage captured the moment an AI pilot takes down three human champions in a fierce drone race.
World drone racing champion Alex Vanover was among three pros beaten by AI as it won an impressive 15 out of 25 trials in a study conducted in Switzerland.
The so-called “Swift” technology mastered speeds of over 100 km/h, with a reaction time an average of 120 milliseconds faster than that of a human.
Its impressive track record is a milestone for University of Zurich scientists trying to push this groundbreaking technology beyond board games.
World drone racing champion Alex Vanover was among three pros beaten by AI as it won an impressive 15 out of 25 trials in a study conducted in Switzerland
“Deep RL (Deep Reinforcement Learning) has enabled some recent advances in artificial intelligence,” scientists say.
“Policies trained with Deep RL have outperformed humans in complex competitive games including Atari, Go, Chess, StarCraft, Dota, and Gran Turismo.”
“Swift went head-to-head against three human champions, including world champions from two international leagues, in real-world head-to-head races.
“This work could inspire the use of hybrid, learning-based solutions in other physical systems such as autonomous ground vehicles, airplanes, and personal robots in a wide range of applications.”
Swift technology uses a complex combination of learning-based algorithms to map a drone’s sensory information while controlling its commands.
As part of that study, pundits pitted Swift against Vanover, as well as two-time MultiGP International Open World Cup winner Thomas Bitmatta and 16-year-old Swiss champion Marvin Schaepper.
The study track featured seven square gates that players had to pass through to complete each 246-foot loop.
To win a race, a competitor had to complete 3 consecutive laps ahead of his opponent in 25 laps.
Each pilot wore a headset that showed a real-time video stream onboard their aircraft, providing a similarly immersive first-person experience.
To win a race in the 25-lap test, a competitor had to complete three consecutive laps ahead of their opponent
The fierce race took place as part of a study developed by the Swiss University of Zurich
Thomas Bitmatta, Marvin Schaepper and Alex Vanover (left to right) compete against KI
The footage shows Swift — lit with blue lights — often hurtling past red human-piloted drones through each gate.
The AI-powered drone outperformed all three champion fliers, posting five, four and six wins over VanoverBitmatta and Schaepper.
Swift also averaged about 16.98 seconds for the three laps – 0.4 seconds faster than second-place Vanover’s 17.38 seconds.
Scientists attribute Swift’s speed to his longer-term thinking style, which helps him make sharper turns and move more efficiently.
On the other hand, it is claimed that human pilots only plan their course of action up to a gate into the future.
Despite this, Swift does have its downsides, as scientists believe it isn’t as resilient as humans.
“Human pilots are impressively resilient: they can crash at full speed and – as long as the hardware is still working – continue to fly and complete the course,” say scientists.
“Swift was not trained to recover from an accident.” “Human pilots are also resilient to changes in environmental conditions, such as lighting, which can significantly change the appearance of the track.”
Still, Swift is a milestone in robotics and experts hope it will inspire other technologies in the future.
“Swift won multiple races against each of the human champions and posted the fastest race time on record,” they wrote.
“This work represents a milestone for mobile robotics and machine intelligence that could inspire the use of hybrid learning-based solutions in other physical systems.”