Robots using BrainBody-LLM can rethink actions mid-task, making them far better at handling complex instructions.

NYU engineers unveil a system that helps robots think through tasks and succeed in settings that usually trip them up.