DKPROMPT: Domain Knowledge Prompting Vision-Language Models for Open-World Planning

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Zhang, Xiaohan, Altaweel, Zainab, Hayamizu, Yohei, Ding, Yan, Amiri, Saeid, Yang, Hao, Kaminski, Andy, Esselink, Chad, Zhang, Shiqi
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909231232319488
author Zhang, Xiaohan
Altaweel, Zainab
Hayamizu, Yohei
Ding, Yan
Amiri, Saeid
Yang, Hao
Kaminski, Andy
Esselink, Chad
Zhang, Shiqi
author_facet Zhang, Xiaohan
Altaweel, Zainab
Hayamizu, Yohei
Ding, Yan
Amiri, Saeid
Yang, Hao
Kaminski, Andy
Esselink, Chad
Zhang, Shiqi
contents Vision-language models (VLMs) have been applied to robot task planning problems, where the robot receives a task in natural language and generates plans based on visual inputs. While current VLMs have demonstrated strong vision-language understanding capabilities, their performance is still far from being satisfactory in planning tasks. At the same time, although classical task planners, such as PDDL-based, are strong in planning for long-horizon tasks, they do not work well in open worlds where unforeseen situations are common. In this paper, we propose a novel task planning and execution framework, called DKPROMPT, which automates VLM prompting using domain knowledge in PDDL for classical planning in open worlds. Results from quantitative experiments show that DKPROMPT outperforms classical planning, pure VLM-based and a few other competitive baselines in task completion rate.
format Preprint
id arxiv_https___arxiv_org_abs_2406_17659
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle DKPROMPT: Domain Knowledge Prompting Vision-Language Models for Open-World Planning
Zhang, Xiaohan
Altaweel, Zainab
Hayamizu, Yohei
Ding, Yan
Amiri, Saeid
Yang, Hao
Kaminski, Andy
Esselink, Chad
Zhang, Shiqi
Artificial Intelligence
Robotics
Vision-language models (VLMs) have been applied to robot task planning problems, where the robot receives a task in natural language and generates plans based on visual inputs. While current VLMs have demonstrated strong vision-language understanding capabilities, their performance is still far from being satisfactory in planning tasks. At the same time, although classical task planners, such as PDDL-based, are strong in planning for long-horizon tasks, they do not work well in open worlds where unforeseen situations are common. In this paper, we propose a novel task planning and execution framework, called DKPROMPT, which automates VLM prompting using domain knowledge in PDDL for classical planning in open worlds. Results from quantitative experiments show that DKPROMPT outperforms classical planning, pure VLM-based and a few other competitive baselines in task completion rate.
title DKPROMPT: Domain Knowledge Prompting Vision-Language Models for Open-World Planning
topic Artificial Intelligence
Robotics
url https://arxiv.org/abs/2406.17659