ZTE Unveils EmbodiedMind Framework for Long-Horizon Robot Planning
让机器人学会「一口气干完一件活」
ZTE Communications has published a paper titled 'EmbodiedMind: Adaptive Data Curation and Prefix-Tree Reinforcement Learning for Efficient Embodied Intelligence,' which introduces a three-stage training framework (RSFT, IR-GRPO, Trie-GRPO) to improve long-horizon planning in embodied intelligence. The 8B model achieved an average score of 70.02% across 18 benchmarks using 522k training samples. The 180B Nebula-EmbodiedBrain model ranked first domestically on the SuperCLUE EmbodiedCLUE-VLA leaderboard in September 2026 with a safety score of 97.92. Performance was validated through simulation and physical tests on AgiBot A2 robots.
Source: Machine Heart — Robotics on WeChat · Read original article ↗
English translation unavailable. Choose the original article to read it.
What the source reports
Publisher-reported claims, with original evidence. These results have not been independently verified by RoboSignal.
Reported numbers
robots
40,000
Reported trials
View original evidence
据中国电子学会数据，2026 年上半年国内人形机器人出货量已超 4 万台，全球占比高达 97%；海外科技巨头亦在加速入场：OpenAI 首席执行官已确认公司将自主研发人形机器人，并直言机器人外观 “远不如真正弄清楚如何让机器人大脑运转重要”，计划 2027 年推出演示样机，核心攻关方向正是驱动机器人运转的智能系统。
Open source S4Reported accuracy
30.31%
View original evidence
在这一背景下，中兴通讯发布论文《EmbodiedMind: Adaptive Data Curation and Prefix-Tree Reinforcement Learning for Efficient Embodied Intelligence》，以 8B 模型在 18 项具身大脑基准上取得 70.02% 的平均成绩，长程规划准确率较现有最优方法提升 30.31%，相关论文已投稿 ICRA 2027。
Open source S9samples
522,000
View original evidence
图 3 Trie-GRPO 自动迭代训练框架（策略与奖励模型联合训练） 从 18 项基准到物理世界，用数据证明 EmbodiedMind-8B 仅使用 52.2 万训练样本，便在 18 项具身大脑基准上取得 70.02% 的平均成绩，综合能力超越同量级开源模型。
Open source S24
- paper: not_reported
What remains unknown
Not established in the collected evidence: Environment, Control, Data origin.
Reported performance applies to the described task. It does not establish general autonomy or deployment readiness.
Source excerpts and review record
Automatically extracted; no manual editorial approval recorded.
据中国电子学会数据，2026 年上半年国内人形机器人出货量已超 4 万台，全球占比高达 97%；海外科技巨头亦在加速入场：OpenAI 首席执行官已确认公司将自主研发人形机器人，并直言机器人外观 “远不如真正弄清楚如何让机器人大脑运转重要”，计划 2027 年推出演示样机，核心攻关方向正是驱动机器人运转的智能系统。
Open source S4
在这一背景下，中兴通讯发布论文《EmbodiedMind: Adaptive Data Curation and Prefix-Tree Reinforcement Learning for Efficient Embodied Intelligence》，以 8B 模型在 18 项具身大脑基准上取得 70.02% 的平均成绩，长程规划准确率较现有最优方法提升 30.31%，相关论文已投稿 ICRA 2027。
Open source S9
随后，SuperCLUE 9 月 EmbodiedCLUE-VLA “具身大脑” 榜单出炉：中兴通讯 Nebula-EmbodiedBrain-180B 以 77.48 分与阿里 Qwen3.8-Max-0902 并列国内第一，具身安全单项高达 97.92 分。
Open source S10
图 3 Trie-GRPO 自动迭代训练框架（策略与奖励模型联合训练） 从 18 项基准到物理世界，用数据证明 EmbodiedMind-8B 仅使用 52.2 万训练样本，便在 18 项具身大脑基准上取得 70.02% 的平均成绩，综合能力超越同量级开源模型。
Open source S24
Source：Machine Heart — Robotics on WeChat · mp.weixin.qq.com