Apple Discloses HALO Autonomous Control System for Humanoid Robots
苹果披露机器人“大脑”研究进展:用的中国大模型
Apple's research team submitted a paper detailing the HALO autonomous control system, validated on Unitree G1 robots to transfer simulation skills to real-world environments. The team released code for data generation, training, and deployment, demonstrating actions like pushing boxes and navigating obstacles. In six unscanned environments, 360 single-task tests achieved an average success rate of 91%; in one environment, three combined task sequences tested 10 times each averaged a 70% success rate.
Source: Chinese robotics — Hardware and sensing · 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
91%
Reported trials
View original evidence
在6个未预先扫描或建图的真实环境中,360次单项任务测试平均成功率为91%;另在一个真实环境中,对3种组合任务分别测试10次,平均成功率为70%,其中“俯身通过障碍—走向目标—推箱子”组合成功率为50%。
Open source S4Reported success rate
70%
Reported trials
View original evidence
在6个未预先扫描或建图的真实环境中,360次单项任务测试平均成功率为91%;另在一个真实环境中,对3种组合任务分别测试10次,平均成功率为70%,其中“俯身通过障碍—走向目标—推箱子”组合成功率为50%。
Open source S4Reported success rate
50%
Reported trials
View original evidence
在6个未预先扫描或建图的真实环境中,360次单项任务测试平均成功率为91%;另在一个真实环境中,对3种组合任务分别测试10次,平均成功率为70%,其中“俯身通过障碍—走向目标—推箱子”组合成功率为50%。
Open source S4test_count
360
Reported trials
View original evidence
在6个未预先扫描或建图的真实环境中,360次单项任务测试平均成功率为91%;另在一个真实环境中,对3种组合任务分别测试10次,平均成功率为70%,其中“俯身通过障碍—走向目标—推箱子”组合成功率为50%。
Open source S4test_count
10
Reported trials
View original evidence
在6个未预先扫描或建图的真实环境中,360次单项任务测试平均成功率为91%;另在一个真实环境中,对3种组合任务分别测试10次,平均成功率为70%,其中“俯身通过障碍—走向目标—推箱子”组合成功率为50%。
Open source S4environment_count
6
View original evidence
在6个未预先扫描或建图的真实环境中,360次单项任务测试平均成功率为91%;另在一个真实环境中,对3种组合任务分别测试10次,平均成功率为70%,其中“俯身通过障碍—走向目标—推箱子”组合成功率为50%。
Open source S4task_combination_count
3
View original evidence
在6个未预先扫描或建图的真实环境中,360次单项任务测试平均成功率为91%;另在一个真实环境中,对3种组合任务分别测试10次,平均成功率为70%,其中“俯身通过障碍—走向目标—推箱子”组合成功率为50%。
Open source S4
- paper: not_reported
- code: Reported pending Open source S3
- dataset: not_reported
- weights: not_reported
- code: 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6个未预先扫描或建图的真实环境中,360次单项任务测试平均成功率为91%;另在一个真实环境中,对3种组合任务分别测试10次,平均成功率为70%,其中“俯身通过障碍—走向目标—推箱子”组合成功率为50%。
Open source S4
苹果披露机器人“大脑”研究进展:用的中国大模型 近日,苹果研究团队提交论文,披露人形机器人自主控制系统HALO,并部署在 宇树科技 G1人形机器人上验证,让机器人将仿真中学到的动作用于陌生现实环境。
Open source S2
团队公开了数据生成、模型训练和部署代码,展示推箱子、跟随人、坐椅子及穿越障碍等动作。
Open source S3
Source:Chinese robotics — Hardware and sensing · taibo.cn