<i>Can an Embodied Agent Find Your “Cat-shaped Mug”?</i> LLM-Based Zero-Shot Object Navigation

摘要

We present language-guided exploration (LGX), a novel algorithm for <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Language-Driven Zero-Shot Object Goal Navigation</i> (L-ZSON), where an embodied agent navigates to an <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">uniquely described</i> target object in a <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">previously unseen</i> environment. Our approach makes use of large language models (LLMs) for this task by leveraging the LLM's commonsense-reasoning capabilities for making sequential navigational decisions. Simultaneously, we perform generalized target object detection using a pre-trained Vision-Language grounding model. We achieve state-of-the-art zero-shot object navigation results on RoboTHOR with a success rate (SR) improvement of over 27% over the current baseline of the OWL-ViT CLIP on Wheels (OWL CoW). Furthermore, we study the usage of LLMs for robot navigation and present an analysis of various prompting strategies affecting the model output. Finally, we showcase the benefits of our approach via <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">real-world</i> experiments that indicate the superior performance of LGX in detecting and navigating to visually unique objects.