At yesterdays meeting, @dcconner gave a great talk on using LTL synthesis to generate FlexBE state machines. He showed how formal specifications can be turned into robust robot behaviors, with fun examples like the classic Farmer’s Dilemma—and shared a demo repo for anyone curious to try it out.
@g-gemignani then kicked off a discussion on how generative AI might help with high-level behavior orchestration. He raised some big-picture questions about combining symbolic planning with perception and learning, sparking an open exchange about where this could lead and what the current limitations are. We are looking into continuing this discussion next month. So if you are interest, please join us.
You can find all references and the plan for next month here: ROS Deliberation Community Group - Google Docs
Christian and @scastro