Inverse Planning

Inverse planning focuses on inferring an agent's goals or intentions from its observed actions and potentially accompanying instructions. Current research emphasizes developing robust algorithms, often employing Bayesian inference and incorporating large language models to handle ambiguity in instructions and multimodal data (actions and language). This field is significant for advancing human-robot collaboration, improving autonomous systems' decision-making (e.g., in robotics and autonomous driving), and providing insights into human social cognition by modeling intention inference.