How close are we to the sci-fi vision of autonomous humanoid robots? I visited 11 companies in five Chinese cities to find out

By Chang Che. Read by Vincent Lai

In feat hailed as milestone in robotics, Sony AI's Ace wins three out of five matches played under official rules An AI-powered robot has beaten elite players at table tennis in a significant achievement for a machine faced with human athletes in a real-world competitive sport. Named Ace, the robotic system developed by Sony AI, won three out of five matches against elite players, but lost the two it played against professionals, clawing back only one game in the seven contests. The feat has been hailed as a milestone for robotics, a field that has long seen table tennis - and the lightning-fast reactions, perception and skill it demands - as one of the toughest tests of how far the technology has advanced. In the matches, played under official competition rules, Ace displayed a mastery of spin, handled difficult shots, such as balls catching on the net, and pulled off one rapid backspin shot that a professional had thought impossible. A research paper on the robot was published in Nature on Wednesday, but scientists working on the project said Ace had improved since the report was submitted.

Mythos: are fears over new AI model panic or PR? - podcast Earlier this month the AI company Anthropic said it had created a model so powerful that, out of a sense of responsibility, it was not going to release it to the public. Anthropic says the model, Mythos Preview, excels at spotting and exploiting vulnerabilities in software, and could pose a severe risk to economies, public safety and national security. But is this the whole story? Some experts have expressed scepticism about the extent of the model's capabilities. Ian Sample hears from Aisha Down, a reporter covering artificial intelligence for the Guardian, to find what the decision to limit access to Mythos reveals about Anthropic's strategy, and whether the model might finally spur more regulation of the industry.

During a summer internship at MIT Lincoln Laboratory, Ivy Mahncke, an undergraduate student of robotics engineering at Olin College of Engineering, took a hands-on approach to testing algorithms for underwater navigation. She first discovered her love for working with underwater robotics as an intern at the Woods Hole Oceanographic Institution in 2024. Drawn by the chance to tackle new problems and cutting-edge algorithm development, Mahncke began an internship with Lincoln Laboratory's Advanced Undersea Systems and Technology Group in 2025. Mahncke spent the summer developing and troubleshooting an algorithm that would help a human diver and robotic vehicle collaboratively navigate underwater. The lack of traditional localization aids -- such as the Global Positioning System, or GPS -- in an underwater environment posed challenges for navigation that Mahncke and her mentors sought to overcome.

The ability to perform effective planning is crucial for building an instructionfollowing agent.

Comprehensiveevaluationson MSQA andMSNN highlight thelimitations ofexisting vision-language models and underscore the importance of handling multi-modal interleaved inputs and situation modeling.

Vision-and-Language Navigation (VLN) aims to develop embodied agents that navigate based on human instructions. However, current VLN frameworks often rely on static environments and optimal expert supervision, limiting their real-world applicability. To address this, we introduce Human-Aware Vision-and-Language Navigation (HA-VLN), extending traditional VLN by incorporating dynamic human activities and relaxing key assumptions.