AI<\/a> models that allow a robot to smoothly complete a wide range of tasks in unfamiliar physical environments remain distant. <\/p>\nAnd as more companies in China’s fledgling humanoid robot industry seek funding and high valuations, investors are placing more importance on real-world deployment over activities such as dance or \u200cathletics to demonstrate their economic value. <\/p>\n
Embodied AI models \u2060determine the intelligence and autonomous operation abilities of robots. Unlike large language models, these physical AI simulation systems are designed to help robots understand and navigate real-world environments in real time. <\/p>\n
Wang Xingxing, the founder of China’s Unitree, this week \u2060predicted that robot brains could see a dramatic breakthrough in two to three years at the earliest. He and other Chinese robotics CEOs have acknowledged that acquiring high-quality real-life training data for these models remains a bottleneck. <\/p>\n
ACE Robotics, which was founded in July 2025 and is backed by Ant Group and SenseTime, has raised more \u200bthan $100 \u200bmillion in the first half of this year through several financing rounds. <\/p>\n
Wang said it \u200baims to launch an initial public offering (IPO) “as early as \u200cpermitted”. Chinese listing rules typically require companies to have at least three fiscal years of operation. <\/p>\n
The company’s open-source Kairos-4B model is ranked first globally by public benchmarks, outperforming world models such as Nvidia’s Cosmos 3 and Ant Group’s Lingbot despite having a much smaller 4 billion parameter count. <\/p>\n
The world model integrates perception, multi-modal understanding, physical simulation and action planning. It can also generate long-horizon video and action predictions over multi-minute sequences. <\/p>\n
EXPANDING TRAINING DATA <\/p>\n
Many robot companies including Unitree and startup X Square are developing their own embodied AI foundation models. <\/p>\n
“Over the past few years, the entire industry has accumulated data of roughly \u200c100,000 hours, which is far from enough to train embodied foundation models,” said ACE’s \u200bWang. <\/p>\n
He said ACE is rapidly scaling data collection by using people on real production lines \u200bfitted with lightweight sensors. <\/p>\n
“The collection efficiency is extremely high … We expect \u200bto accumulate tens of millions of hours of data within two years,” Wang added. <\/p>\n
Many firms train humanoid robots using \u200cphysical tele-operation, where workers wearing exoskeletons and controllers repeat simple \u200bphysical movements hundreds of times a day. <\/p>\n
ACE \u200bis deploying its embodied AI models in commercial settings such as unmanned retail stores, hotel services and instant-delivery warehouses staffed by humanoid robots from Chinese makers including Unitree, AgiBot and Fourier. <\/p>\n
“We plan to deploy in at least 1,000 stores over the coming year, scaling to \u200b10,000 stores in two years,” said Wang. <\/p>\n
For training its \u200cmodels, ACE uses AI chips from Nvidia<\/a> as well as Chinese makers including Rhino Tech and Digua Robotics. <\/p>\n“This diversifies our supply \u200bchain, because in the future, we will definitely need comprehensive intelligent hardware solutions, and their costs need to be significantly \u200breduced,” said Wang. <\/p>\n
(Reporting by Laurie Chen; Editing by Eduardo Baptista and Alexander Smith) <\/p>\n","protected":false},"excerpt":{"rendered":"By Laurie Chen BEIJING, Aug 21 (Reuters) – Humanoid robot brains could see a breakthrough by late next…\n","protected":false},"author":2,"featured_media":147214,"comment_status":"","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[7],"tags":[580,29974,34011,1834,4436,157,6196,377,36382,71678,10857],"class_list":["post-147213","post","type-post","status-publish","format-standard","has-post-thumbnail","category-openai","tag-chatgpt","tag-data-capture","tag-embodied-intelligence","tag-humanoid-robot","tag-language-models","tag-openai","tag-robot","tag-robotics","tag-wang-xiaogang","tag-wang-xingxing","tag-world-models"],"_links":{"self":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/posts\/147213","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/comments?post=147213"}],"version-history":[{"count":0,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/posts\/147213\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/media\/147214"}],"wp:attachment":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/media?parent=147213"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/categories?post=147213"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/tags?post=147213"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}