Chinese Humanoid Robot T800 Knocks Out Human Boxer in SF Bout

In a robot combat event that blurred the line between spectacle and hardware R&D, a Chinese-built humanoid robot took down a human boxer in San Francisco. As reported by Korea's 로봇신문 (Robot Newspaper), citing the Chinese robotics outlet 중국로봇망 (China Robot Net), the humanoid "T800" built by Chinese startup EngineAI (众擎机器人) defeated boxer Frankie LaPenna with a powerful kick during the REK robot combat competition.

An important caveat: this was not a fully autonomous AI fight. The robot was remotely piloted by an operator wearing a VR headset and motion controllers, while the robot itself independently handled dynamic balance and posture correction — a hybrid setup combining human teleoperation with low-level motion control running on the robot.

For this particular bout, the standard T800 was modified to stand 1.83 meters tall, 10 centimeters taller than the stock version, and outfitted with armor styled after the Terminator films for a suitably intimidating look.

T800 is EngineAI's full-size, high-performance humanoid platform, priced starting at 180,000 yuan (roughly 36.5 million KRW). It weighs 75 kg and can deliver up to 450 Nm of torque per joint. The robot runs on an in-house solid-state battery that provides 4 to 5 hours of continuous operation per charge, and its dozens of flexible joints let it run, throw spinning kicks, and recover its footing after falling.

This wasn't T800's combat-ring debut. At an earlier event — the URKL humanoid robot fighting competition held in China — the robot kept throwing punches even after an opponent's kick knocked its head clean off, a moment that drew wide attention to the resilience of its sensors and control algorithms. In a separate incident, EngineAI's own CEO, Zhao Tongyang, reportedly was knocked down while personally testing the robot.

EngineAI says the purpose of entering sports competitions is to gather data from high-intensity collision scenarios, which the company hopes will help refine the robot's balance, force control, and perception algorithms. The ultimate goal, according to the company, is deployment in industrial and service-sector applications rather than the fighting ring itself.