Skip to content
Chinese robotics — Hardware and sensing·· 20 hours agoSignalEditorial score85

Humanoid Robots Are About to Deliver Big Results

人形机器人,要出大结果了!

Summary

The article discusses the rapid development of humanoid robots, with Tesla and NVIDIA leading the charge in production and simulation. It outlines the technical challenges of physical AI, including the need for real-world data and the role of simulation platforms like NVIDIA's Isaac. The piece also discusses the competitive landscape, with Chinese companies like Zhuyu and Zhixun leading in production, and the potential for physical AI to transform manufacturing and logistics.

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

  • robot_trajectories

    780,000

    Trajectory count

    View original evidence
    英伟达去年3月公布过一组数据:用它的工具,11个小时就能生成78万条机器人合成动作轨迹,相当于人类示范6500个小时的数据量。
    Open source S17
  • robot_trajectories

    6,500 hours

    View original evidence
    英伟达去年3月公布过一组数据:用它的工具,11个小时就能生成78万条机器人合成动作轨迹,相当于人类示范6500个小时的数据量。
    Open source S17
  • Reported success rate

    100%

    View original evidence
    宇树科技创始人王兴兴8月在世界机器人大会上就讲过:在固定场景里,经过充分的数据采集和专项训练,机器人完成任务的成功率几乎能到100%;可只要换一件物品,或者把环境稍微挪一挪,成功率就会大幅下滑。
    Open source S18
  • robot_cost

    10,000

    View original evidence
    早在2024年6月的特斯拉股东大会上,他就算过一笔大账:全世界未来每年可能要造10亿台人形机器人,特斯拉的份额可能远超10%;规模足够大时,每台成本可降到1万美元左右,卖2万美元。
    Open source S21
  • robot_cost

    20,000

    View original evidence
    早在2024年6月的特斯拉股东大会上,他就算过一笔大账:全世界未来每年可能要造10亿台人形机器人,特斯拉的份额可能远超10%;规模足够大时,每台成本可降到1万美元左右,卖2万美元。
    Open source S21
  • robot_production

    1,000,000,000 robots

    View original evidence
    早在2024年6月的特斯拉股东大会上,他就算过一笔大账:全世界未来每年可能要造10亿台人形机器人,特斯拉的份额可能远超10%;规模足够大时,每台成本可降到1万美元左右,卖2万美元。
    Open source S21
  • paper: 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.

英伟达去年3月公布过一组数据:用它的工具,11个小时就能生成78万条机器人合成动作轨迹,相当于人类示范6500个小时的数据量。

Open source S17

王兴兴给具身智能的“ChatGPT时刻”下的标准是:把一台机器人带到任意一个陌生环境,比如你家里,它基本能完成80%左右的任务。

Open source S19

宇树科技创始人王兴兴8月在世界机器人大会上就讲过:在固定场景里,经过充分的数据采集和专项训练,机器人完成任务的成功率几乎能到100%;可只要换一件物品,或者把环境稍微挪一挪,成功率就会大幅下滑。

Open source S18

早在2024年6月的特斯拉股东大会上,他就算过一笔大账:全世界未来每年可能要造10亿台人形机器人,特斯拉的份额可能远超10%;规模足够大时,每台成本可降到1万美元左右,卖2万美元。

Open source S21

一台人形机器人通常有四十个左右的关节,每个关节里塞着电机、减速器、丝杠和传感器。

Open source S11

Source:Chinese robotics — Hardware and sensing · sohu.com

Timezone · UTC

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