IROS 2023 Late Breaking Results (Oral)
We not only shown a poster of our work, but also had a chance to present our work in formal session.
Last updated
We not only shown a poster of our work, but also had a chance to present our work in formal session.
Last updated
Sikai Li, Run Peng, Joyce Chai, "GPT-4 Empowered Theory of Mind Modeling for Human-Robot Collaboration", in IROS Late Breaking Results, 2023.
Humans develop Theory of Mind (ToM) at a young age - the ability to understand that others may have intents, beliefs, knowledge, skills, etc. that may differ from our own. Modeling others' belief states, in other words, ToM, plays an important role in human-human communication and collaborative tasks. As a new generation of cognitive robots start to enter our lives, it's important for these robots to have similar ToM ability in order to effectively collaborate with humans. While there is an increasing amount of work in ToM modeling for collaborative tasks in human-agent collaboration, most of the works were situated in a simulated environment. In this work, we investigate ToM modeling in human-robot communication and collaboration. As large language models have shown impressive abilities in language communication with humans, we particularly incorporate the latest advances of GPT-4 in modeling intentions of humans for efficient decision making in collaborative tasks with humans. We mainly focus on the prediction of human's actions based on their intentions during collaboration. Leveraging the advanced reasoning ability of large language model GPT-4, ToM can be modeled by prompt engineering under zero-shot settings.
