Apple Research Team Discloses Progress on Humanoid Robot 'Brain' System HALO
苹果研究团队披露机器人“大脑”进展：单项任务测试平均成功率为91%
On November 4, Apple's research team submitted a preprint paper titled 'Perceptive Humanoid Motion from Simulation' detailing the autonomous control system HALO. The team released code for action generation, tracking, and task orchestration. Tests in six unscanned environments yielded an average single-task success rate of 91%, while combined tasks averaged 70%. Training utilized NVIDIA H100 and B200 GPUs, though data synthesis involved human prompts and motion capture. The system currently relies on visual perception, lacking direct contact sensing.
Source: Chinese robotics — Research news discovery · 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.
- Control
- Reported autonomousOpen source S3
Reported numbers
Reported success rate
91%
Unique elapsed hours
Reported success rate
70%
Reported success rate
50%
test_count
360
Reported trials
test_count
10
Reported trials
gpu_hours
≈48 hours
- paper: not_reported
- code: Reported pending Open source S4
- weights: not_reported
- dataset: not_reported
- code: not_reported
What remains unknown
Not established in the collected evidence: Environment, 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.
据论文披露，研究团队在6个未预先扫描或建图的真实环境中开展360次单项任务测试，平均成功率为91%。
Open source S5
当地时间11月4日，苹果研究团队在预印本平台arXiv提交论文《Perceptive Humanoid Motion from Simulation》，披露了人形机器人自主控制系统HALO。
Open source S3
另在一个真实环境中，对3种组合任务分别开展10次测试，平均成功率为70%。
Open source S6
其中，“俯身通过障碍—走向目标—推箱子”这一组合的成功率为50%。
Open source S7
据介绍，研究团队使用250张英伟达H100 GPU，耗时约48小时合成主要训练数据。
Open source S8
动作生成器通过模仿学习训练，使用64张B200；动作跟踪器经强化学习、策略蒸馏等步骤训练，使用4张H100。
Open source S9
研究团队还公开了数据生成、模型训练和部署代码，涵盖动作生成、动作跟踪、任务编排以及仿真和真实机器人运行等环节。
Open source S4
Source：Chinese robotics — Research news discovery · sohu.com