Humanoid robots compete in marathon in Beijing

Image Source: Unsplash. Representative

In what is very likely one of the first-ever robot marathons, 21 humanoid robots took part in a half-marathon alongside thousands of human runners in Beijing's E-Town tech hub on Saturday — though most mechanical competitors staggered to the finish line well behind their human counterparts.

The winning robot, Tiangong Ultra, developed by the government-backed research institute X-Humanoid, completed the 21-kilometre (13.1-mile) race in 2 hours and 40 minutes — a respectable effort for a robot, but far slower than the winning male human runner, who clocked in at 1 hour and 2 minutes. 

Tiangong Ultra required a human assistant to lead the way, using a signaling device worn on his back to help the robot mimic his movements. Other robots relied on remote control systems, with human operators running alongside them to guide their progress.

According to TechCrunch, only four robots managed to cross the finish line within the four-hour cutoff, with many others struggling to even get moving. 

One robot, Shennong, tripped its support runner, crashed into a fence, and shattered shortly after the race began. Another, Little Giant, came to a dramatic halt when smoke started emerging from its head.

Despite setbacks, the Beijing E-Town Humanoid Robot Half Marathon marked a milestone in robotics and athletic engineering, with entries from major Chinese companies and student groups. 

Robots were required to have a humanoid appearance and run on two legs. They ran in a fenced-off lane with staggered start times. Battery swaps were allowed — Tiangong Ultra's battery was changed three times — and substitute robots could be swapped in, with a time penalty.

"This is just the beginning," Tang Jiang, Chief Technology Officer of X-Humanoid, told Reuters. "I don't want to boast but I think no other robotics firms in the West have matched Tiangong’s sporting achievements."

Organisers hailed the event as a glimpse into the future of robotics and physical AI performance.