End-to-End Navigation with Vision Language Models: Transforming Spatial Reasoning into Question-Answering

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Goetting, Dylan, Singh, Himanshu Gaurav, Loquercio, Antonio
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910690233548800
author Goetting, Dylan
Singh, Himanshu Gaurav
Loquercio, Antonio
author_facet Goetting, Dylan
Singh, Himanshu Gaurav
Loquercio, Antonio
contents We present VLMnav, an embodied framework to transform a Vision-Language Model (VLM) into an end-to-end navigation policy. In contrast to prior work, we do not rely on a separation between perception, planning, and control; instead, we use a VLM to directly select actions in one step. Surprisingly, we find that a VLM can be used as an end-to-end policy zero-shot, i.e., without any fine-tuning or exposure to navigation data. This makes our approach open-ended and generalizable to any downstream navigation task. We run an extensive study to evaluate the performance of our approach in comparison to baseline prompting methods. In addition, we perform a design analysis to understand the most impactful design decisions. Visual examples and code for our project can be found at https://jirl-upenn.github.io/VLMnav/
format Preprint
id arxiv_https___arxiv_org_abs_2411_05755
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle End-to-End Navigation with Vision Language Models: Transforming Spatial Reasoning into Question-Answering
Goetting, Dylan
Singh, Himanshu Gaurav
Loquercio, Antonio
Robotics
Computation and Language
Computer Vision and Pattern Recognition
We present VLMnav, an embodied framework to transform a Vision-Language Model (VLM) into an end-to-end navigation policy. In contrast to prior work, we do not rely on a separation between perception, planning, and control; instead, we use a VLM to directly select actions in one step. Surprisingly, we find that a VLM can be used as an end-to-end policy zero-shot, i.e., without any fine-tuning or exposure to navigation data. This makes our approach open-ended and generalizable to any downstream navigation task. We run an extensive study to evaluate the performance of our approach in comparison to baseline prompting methods. In addition, we perform a design analysis to understand the most impactful design decisions. Visual examples and code for our project can be found at https://jirl-upenn.github.io/VLMnav/
title End-to-End Navigation with Vision Language Models: Transforming Spatial Reasoning into Question-Answering
topic Robotics
Computation and Language
Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2411.05755