At UC Berkeley, scientists in Sergey Levine’s Robot AI and Understanding Laboratory looked at a table where a tower of 39 Jenga obstructs stood completely piled. After that a white-and-black robotic, its solitary arm or leg folded like a hunched-over giraffe, zoomed towards the tower, sporting a black natural leather whip. Via what could have appeared to an informal audience like a wonder of physics, the whip struck in exactly the appropriate area to send out a solitary block flying from the pile while the remainder of the tower stayed structurally audio.
发布者：Dr.Durant，转转请注明出处：https://robotalks.cn/with-human-feedback-ai-driven-robots-learn-tasks-better-and-faster/
