Gaode Releases First Full-Body Base Model λ0 Based on Open-World Behavior
高德发布首个基于开放世界行为的全身基座模型λ0，推动机器人从“局部学习”迈向“全身动作对齐”
Gaode, under the Alibaba Group, officially released its first full-body base model λ0 based on open-world behavior. This model enables robots to learn full-body coordination, including movement, posture, and hand operations, solely through human action data, thus overcoming the limitations of 'local action learning.' In the SIMPLE simulation benchmark and multiple real-world evaluations, λ0 achieved state-of-the-art performance, with data collection efficiency approximately 3.4 times that of mainstream methods and a 22.5% increase in real-world task success rate compared to the strongest baseline.
The release of λ0 represents a significant step in embodied AI by enabling full-body coordination in robots through human action data, without requiring specialized hardware. The model's performance on both simulation and real-world benchmarks highlights its potential for broader application in open environments.
Source: QbitAI — Robotics · 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
data_collection_efficiency
3.4
View original evidence
在SIMPLE仿真基准及多项真机评测中，λ0均取得SOTA表现，数据采集效率约为主流方法的3.4倍，真机任务成功率较最强对比基线提升22.5%。
Open source S4Reported success rate
22.5%
View original evidence
在SIMPLE仿真基准及多项真机评测中，λ0均取得SOTA表现，数据采集效率约为主流方法的3.4倍，真机任务成功率较最强对比基线提升22.5%。
Open source S4single_demo_collection_time
29 seconds
data_collection_efficiency_comparison
3.4
data_set_size
500 hours
- dataset: Available Artifact link Open source S22
- paper: Available Artifact link Open source S22
- paper: not_reported
- dataset: 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.
在SIMPLE仿真基准及多项真机评测中，λ0均取得SOTA表现，数据采集效率约为主流方法的3.4倍，真机任务成功率较最强对比基线提升22.5%。
Open source S4
与机器人本体采集相比，该方式将代表性任务的单条示范采集用时由约100秒缩短至29秒，采集效率约为原来的3.4倍。
Open source S11
基于这一可规模化的管线，高德构建了包含500小时开放世界无本体全身数据的HumanVerse-500数据集。
Open source S13
同时，在同等机器人示范量的条件下，随着人类全身数据规模的扩大，真实机器人的任务表现能获得显著提升。
Open source S16
这意味着，规模化积累人类全身数据，是提升人形机器人真实任务表现的一条有效路径。
Open source S17
在跨形态迁移与数据扩展性（Scaling）方面，λ0面对7项全身协调与长程移动操作任务，仅凭人类行动数据即可在真机上实现零样本迁移。
Open source S15
附 项目主页：https://lambdahumanoid.github.io/ 技术报告：https://arxiv.org/abs/2610.18950 GitHub：https://github.com/LambdaHumanoid/openlambda HumanVerse 数据样例：https://huggingface.co/datasets/lambda-humanoid/HumanVerse-500-sample 本文由高德提供，量子位获授权转载，观点归原作者所有。
Open source S22
Source：QbitAI — Robotics · qbitai.com