ICRA 2026 现场：中国军团主导维也纳，新锐巨头混战具身智能深水区
Quick Answer
At ICRA 2026 in Vienna, Chinese companies dominated with significant advancements in embodied intelligence, showcasing models like Spirit v1.6 from Qianxun Technology, which outperformed NVIDIA's Cosmos3.
Quick Take
The event highlighted a shift towards physical intelligence, with 4947 submissions and a 38.04% acceptance rate, emphasizing China's rising influence in the global robotics landscape.
Key Points
- ICRA 2026 featured over 8000 attendees from 86 countries, emphasizing global collaboration.
- Chinese universities accounted for 55% of the top 20 accepted papers, with Tsinghua University leading at 74.
- The event showcased 153 workshops covering cutting-edge topics like embodied intelligence and Sim-to-Real.
- Qianxun Technology's Spirit v1.6 model ranked first in global benchmarks against NVIDIA's latest models.
- Significant IPOs and partnerships, including Yushutech's rapid IPO approval, marked the event's momentum.
Source Excerpt
史上最大ICRA，中国惊艳世界。 作者丨吴思梦 编辑丨岑 峰 宇树科技、千寻智能、帕西尼、智元机器人、加速进化、高擎机电、光轮智能、银河通用、魔法原子、戴盟机器人、鹿明机器人等企业在本届 ICRA 的核心工业展区，放眼望去皆是中国展商的身影。 这一幕，让日本参会学者Shohei Hido在现场直接感叹：“China is dominating”——中国展商，已经成为本届ICRA最醒目的存在。 当多瑙河畔的音乐之都响起代码与算法的交响，全球机器人学界一年一度的“奥林匹克”——IEEE国际机器人与自动化会议（ICRA 2026），于6月1日在奥地利维也纳正式拉开帷幕。 本届大会以“Robots for All”为主题，于6月1日至5日举行，汇聚了来自86个国家和地区的超过8000名学者与产业人士。 雷峰网已抵达第一现场，从组委会获取的最终数据显示：本届ICRA有效投稿达4947篇，最终录用1882篇，录用率为38. 04%其中，11所中国高校占Top 20的55%，清华74篇断层第一，"清北交浙"四校合计212篇。 三年间投稿量增幅超过50%，门槛不降反升。
大会共设有153场Workshop，覆盖具身智能、VLA、Sim-to-Real等前沿方向，另有20项竞赛提案同台竞技。 如果说去年的关键词还是“大模型”，那么今年ICRA的核心叙事已明确转向“物理智能”——而在这场从感知到行动的范式转移中，中国企业正在以前所未有的力度改写全球机器人产业版图。 …
Want this in your inbox every morning?
Daily brief at your local 8am — bilingual EN/中文, free.
More from 雷峰网机器人
See more →
WAIC深度观察：具身数据，进入「开箱即用」时代
The WAIC forum highlights the urgent need for 'Model-Ready Data' in embodied intelligence, emphasizing high precision, multi-modality, and diversity as essential for effective physical interaction. Companies like JianZhi Robotics are pioneering solutions to bridge existing data gaps, ensuring that data is directly usable without further processing.
