HARBOR builds robot policies from one prompt

Georgia Chalvatzaki, a professor of embodied AI at TU Darmstadt, retweeted a post by @softraeh. The post states that one prompt and one simulator allow HARBOR to autonomously build the task, design rewards, train, and tune a robot policy. The visible source line contains only this claim and the retweet action. No additional posts, confirmations, or outcomes appear in the packet.