Do As I Can, Not As I Say: Grounding Language in Robotic Affordances

TLDR

Large language models encode rich semantic knowledge but lack real‑world experience, limiting their usefulness for embodied robots executing natural‑language instructions. We aim to ground language models in real‑world affordances by constraining them with pretrained robotic skills. The robot acts as the language model’s hands and eyes, and pretrained skills constrain the model to generate feasible, context‑appropriate natural‑language actions. Our experiments on a mobile manipulator demonstrate that this grounding enables the system to complete long‑horizon, abstract natural‑language tasks, confirming the necessity of real‑world grounding. The project website and video are available at https://say‑can.github.io/.

Abstract

Large language models can encode a wealth of semantic knowledge about the world. Such knowledge could be extremely useful to robots aiming to act upon high-level, temporally extended instructions expressed in natural language. However, a significant weakness of language models is that they lack real-world experience, which makes it difficult to leverage them for decision making within a given embodiment. For example, asking a language model to describe how to clean a spill might result in a reasonable narrative, but it may not be applicable to a particular agent, such as a robot, that needs to perform this task in a particular environment. We propose to provide real-world grounding by means of pretrained skills, which are used to constrain the model to propose natural language actions that are both feasible and contextually appropriate. The robot can act as the language model's "hands and eyes," while the language model supplies high-level semantic knowledge about the task. We show how low-level skills can be combined with large language models so that the language model provides high-level knowledge about the procedures for performing complex and temporally-extended instructions, while value functions associated with these skills provide the grounding necessary to connect this knowledge to a particular physical environment. We evaluate our method on a number of real-world robotic tasks, where we show the need for real-world grounding and that this approach is capable of completing long-horizon, abstract, natural language instructions on a mobile manipulator. The project's website and the video can be found at https://say-can.github.io/.