In a piece for The Conversation, republished by Robohub, Wai-keung Fung of Cardiff Metropolitan University argues that athletic feats say little about whether humanoid robots are ready for everyday life. At the World Humanoid Robot Games 2026 in Beijing, robots sprinted, boxed and played football, and the robot Tiangong Ultra completed 100 metres in 8.64 seconds, faster than Usain Bolt's 9.58-second world record.
Fung writes that no robot system can yet reliably handle the whole laundry process, from carrying a basket to the machine and sorting clothes to drying, folding and ironing. A sprint is a well-defined task on a predictable track; fabric is soft and constantly changing shape, shirts may be inside out and sleeves hidden under other garments. A useful household robot, he argues, must notice when reality differs from what it expected and adapt, especially in homes where objects, people and pets move unpredictably.
Physical skills cannot simply be installed on any machine like an app. A specialist laundry robot with rigid arms and fabric grippers differs from a humanoid with head cameras, five-fingered hands and balance constraints, so the same movement can produce different results. Some knowledge does transfer, such as what a good fold looks like, but each robot must work out how to execute it with its own body and sensors.
Research projects such as Open X-Embodiment are tackling this by pooling experience from many robots. According to the article, the project has brought together more than one million real robot trajectories from 22 different robot types, aiming to let robots learn from one another rather than starting from zero.
Fung proposes questions closer to real usefulness: can a robot handle an unfamiliar object, notice its own mistake and recover without human help, and how long can it work before someone has to step in? These measures are unlikely to go viral, he writes, but they will determine whether general-purpose robot intelligence has really arrived.