SG-CoT: An Ambiguity-Aware Robotic Planning Framework using Scene Graph Representations

TL;DR

SG-CoT is a novel framework that enhances robotic planning by using scene graph representations and iterative language model querying to resolve ambiguities, leading to more reliable and generalizable robot behavior.

Contribution

It introduces a two-stage scene graph-based framework enabling LLMs to detect, clarify, and resolve ambiguities in robotic planning tasks.

Findings

10%+ improvement in question accuracy

4%+ success rate increase in single-agent environments

15%+ success rate increase in multi-agent environments

Abstract

Ambiguity poses a major challenge to large language models (LLMs) used as robotic planners. In this letter, we present Scene Graph-Chain-of-Thought (SG-CoT), a two-stage framework where LLMs iteratively query a scene graph representation of the environment to detect and clarify ambiguities. First, a structured scene graph representation of the environment is constructed from input observations, capturing objects, their attributes, and relationships with other objects. Second, the LLM is equipped with retrieval functions to query portions of the scene graph that are relevant to the provided instruction. This grounds the reasoning process of the LLM in the observation, increasing the reliability of robotic planners under ambiguous situations. SG-CoT also allows the LLM to identify the source of ambiguity and pose a relevant disambiguation question to the user or another robot. Extensive experimentation demonstrates that SG-CoT consistently outperforms prior methods, with a minimum of 10% improvement in question accuracy and a minimum success rate increase of 4% in single-agent and 15% in multi-agent environments, validating its effectiveness for more generalizable robot planning.