Artificial Intelligence Beats Humans in Drone Obstacle Race

In a groundbreaking study in the field of mobile robotics, an AI-assisted drone has outperformed a human-controlled drone in obstacle flight tests, demonstrating more precise flight patterns and speed. AI-assisted machines have already defeated humans in non-sporting games such as chess, checkers, Go, Othello, and StarCraft. However, researchers at the University of Zurich witnessed for the first time a drone achieving victory in a physical challenge. With the emergence of smaller, faster drones equipped with high-resolution cameras, first-person view (FPV) flying has become increasingly popular. In an article published in Science Robotics, the team from the University of Zurich reported that their device, named "Swift," challenged three expert drone pilots and defeated them 15 times out of 25 challenges. Elia Kaufmann, a member of the team, stated, "Our results mark the first time that a robot has defeated a human champion in a physically designed game for humans." The Swift system relies on an artificial neural network to optimize the drone's flight path and speed. It collects real-time details through an onboard camera, while human pilots rely on video transmitted to a head-mounted display. The competition consisted of a course with seven square pillars that the drone had to navigate across a field of approximately 27 meters. The goal was to complete the task in the shortest time possible without any accidents. In addition to defeating human pilots in dozens of trials, the Swift drone also set a record for the fastest speed, albeit only half a second faster. According to the article, "This task requires the drone to maintain its speed and acceleration at the physical limits, with low tolerance for errors: a small mistake could lead to a catastrophic crash or significantly impact lap times." The researchers noted that Swift's efficiency decreases when minor changes, such as lighting, occur. Intelligent drones have enormous potential in various fields, not just for delivering Amazon's latest products. Farmers can use drones to monitor crop growth, diseases, and pests. Engineers can employ drones to inspect collapsed power lines or damaged bridges, tasks that are dangerous for humans but must be completed. Drones can be deployed in search and rescue operations during natural disasters in areas of rubble or flooding. Police can launch drones to assist in apprehending fugitives. In the realm of sports, drones can provide previously unimaginable aerial filming perspectives. The PGA Tour also utilizes drone photography technology. Currently, drones have not yet participated in such sporting events. However, when they do appear on the field, human players will need to take notice.