World Models × Embodied Intelligence | 2026 Five Representative Companies' Technical Paths and Scene Deployment Observations
世界模型×具身智能｜2026年五家代表企业技术路径与场景落地观察
This article observes the technical paths and scene deployment progress of five Chinese companies in the field of world models and embodied intelligence. It evaluates their core competencies and development prospects, emphasizing the transition from parameter scale to system-level performance and the importance of real-world data feedback in model iteration.
The article provides an observational analysis of five Chinese robotics companies' approaches to world models and embodied intelligence, focusing on their technical strategies, deployment capabilities, and industry impact. It highlights the shift from parameter scale to system-level performance and the importance of real-world data feedback in model iteration.
Source: Chinese robotics — Robot learning research · 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
Reported success rate
98.6%
View original evidence
公开评测中，ω-EVA在LIBERO达到98.6%成功率，LIBERO-PLUS达到83.4%，RoboTwin达到90.3%。
Open source S11parameters in JEPA-WAM
0.5
View original evidence
JEPA-WAM采用Shared Predictor共享预测器，学习跨任务环境时序变化规律，基座为0.5B参数，无需大规模机器人预训练数据和像素级未来视频生成，相关代码已开源。
Open source S12calculation power of T5 and N5
2,070
- paper: not_reported
- code: not_reported
- dataset: not_reported
- code: Reported pending Open source S12
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.
公开评测中，ω-EVA在LIBERO达到98.6%成功率，LIBERO-PLUS达到83.4%，RoboTwin达到90.3%。
Open source S11
JEPA-WAM采用Shared Predictor共享预测器，学习跨任务环境时序变化规律，基座为0.5B参数，无需大规模机器人预训练数据和像素级未来视频生成，相关代码已开源。
Open source S12
T5与N5搭载Jetson Thor T5000，提供2070 TFLOPS算力，已在智元机器人精灵G2上批量出货。
Open source S18
BotPack B系列算力背包面向存量机器人高算力扩展，已在宇树G1等机器人上实现即插即用。
Open source S19
（三）具身大脑与跨本体调度 RoboBrain Pro具身大脑提供多模态感知、场景理解、任务规划与运动控制能力，可适配人形双足、轮式机器人、机械臂、机器狗、特种机器人、叉车等形态。
Open source S14
RoboBrain Xross跨本体具身大脑调度平台通过全局感知、任务编排、动态调度，打通不同本体的信息通路与决策链路，实现“一脑调度多台机器人”。
Open source S15
Source：Chinese robotics — Robot learning research · news.zol.com.cn