Hao AI Lab· @haoailab · X·· 2 小时前AI 评分43
基准测试为何测不出机器人真实延迟
AI 导读
为什么基准测试显示不出这一点?在 RoboDojo 中,仿真器会等模型完成再执行下一步,但真实机器人不会等。 我们早期叠毛巾的运行耗时 13 到 37 分钟。在其中一次最长的运行中,模型占了 78% 的实际耗时,而机器人只运行了全程的四分之一。
正文 · 原文
Why don't benchmarks show this? In RoboDojo, the simulator waits for the model before it steps, but a real robot doesn't wait.
Our early towel-folding runs took 13 to 37 minutes. In one of the longest, the model took 78% of the wall time, and the robot moved for only a quarter of the run.
来源:Hao AI Lab · x.com