Skip to content
Chinese robotics — Hardware and sensing·· 6 days agoSignalEditorial score75

Apple Research Team Unveils HALO Autonomous Control System for Humanoid Robots

91%成功率!苹果亮出人形机器人大脑HALO 陌生环境直接上手

Summary

Apple's research team recently published findings on its HALO autonomous control system for humanoid robots, which uses visual-language models and action generation to understand human instructions and adapt movements in unfamiliar environments. The system was validated on Unitree G1 robots across six unscanned real-world environments, achieving an average success rate of 91% for single tasks and 70% for multi-step tasks. Training involved generating data in 1,000 simulated home environments using 250 NVIDIA H100 GPUs.

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.

Control
Reported autonomousOpen source S3

Reported numbers

  • Reported success rate

    91%

    Reported trials

    View original evidence
    在实际测试方面,研究团队选择6个未经事先扫描或建图的真实环境,进行360次单项任务测试,平均成功率达91%。
    Open source S17
  • Reported success rate

    70%

    Reported trials

    View original evidence
    研究团队另外在单一真实环境中,针对3种复合任务各进行10次测试,平均成功率为70%。
    Open source S18
  • Reported success rate

    50%

    Reported trials

    View original evidence
    其中,“弯腰通过障碍物、走向目标、推动箱子”这项连续任务的成功率仅50%,显示机器人处理多步骤动作时仍面临挑战。
    Open source S19
  • training_hours

    ≈300 hours

    View original evidence
    为了训练机器人适应不同环境,研究团队建立一套大规模合成资料生成流程,在1,000个程序生成的模拟家庭环境中,制作约300小时的主要训练资料,将机器人看到的画面、接收的指令及对应动作配对。
    Open source S20
  • gpu_hours

    ≈48 hours

    View original evidence
    在计算资源方面,研究团队使用250张英伟达H100 GPU,耗时约48小时完成主要训练资料的合成;动作生成器采用模仿学习训练,使用64张B200 GPU;动作追踪器则通过强化学习、策略蒸馏等方式训练,使用4张H100 GPU。
    Open source S23
  • trial_count

    360

    Reported trials

    View original evidence
    在实际测试方面,研究团队选择6个未经事先扫描或建图的真实环境,进行360次单项任务测试,平均成功率达91%。
    Open source S17
  • trial_count

    10

    Reported trials

    View original evidence
    研究团队另外在单一真实环境中,针对3种复合任务各进行10次测试,平均成功率为70%。
    Open source S18
  • simulation_environment_count

    1,000

    View original evidence
    为了训练机器人适应不同环境,研究团队建立一套大规模合成资料生成流程,在1,000个程序生成的模拟家庭环境中,制作约300小时的主要训练资料,将机器人看到的画面、接收的指令及对应动作配对。
    Open source S20
  • paper: not_reported
  • code: not_reported
  • weights: not_reported
  • dataset: 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.

该系统通过视觉语言模型(VLM)与动作生成技术,让机器人不仅能理解人类指令,还能依据陌生环境调整动作,自主完成推箱子、坐椅子及穿越障碍物等任务。

Open source S3

在实际测试方面,研究团队选择6个未经事先扫描或建图的真实环境,进行360次单项任务测试,平均成功率达91%。

Open source S17

研究团队另外在单一真实环境中,针对3种复合任务各进行10次测试,平均成功率为70%。

Open source S18

其中,“弯腰通过障碍物、走向目标、推动箱子”这项连续任务的成功率仅50%,显示机器人处理多步骤动作时仍面临挑战。

Open source S19

为了训练机器人适应不同环境,研究团队建立一套大规模合成资料生成流程,在1,000个程序生成的模拟家庭环境中,制作约300小时的主要训练资料,将机器人看到的画面、接收的指令及对应动作配对。

Open source S20

在计算资源方面,研究团队使用250张英伟达H100 GPU,耗时约48小时完成主要训练资料的合成;动作生成器采用模仿学习训练,使用64张B200 GPU;动作追踪器则通过强化学习、策略蒸馏等方式训练,使用4张H100 GPU。

Open source S23

Source:Chinese robotics — Hardware and sensing · finance.sina.com.cn

Timezone · UTC

Article dates follow your selected timezone. Briefing editions use Hong Kong time (UTC+8).