X-MOBILITY: End-To-End Generalizable Navigation via World Modeling

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Liu, Wei, Zhao, Huihua, Li, Chenran, Biswas, Joydeep, Okal, Billy, Goyal, Pulkit, Chang, Yan, Pouya, Soha
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866912497586405376
author Liu, Wei
Zhao, Huihua
Li, Chenran
Biswas, Joydeep
Okal, Billy
Goyal, Pulkit
Chang, Yan
Pouya, Soha
author_facet Liu, Wei
Zhao, Huihua
Li, Chenran
Biswas, Joydeep
Okal, Billy
Goyal, Pulkit
Chang, Yan
Pouya, Soha
contents General-purpose navigation in challenging environments remains a significant problem in robotics, with current state-of-the-art approaches facing myriad limitations. Classical approaches struggle with cluttered settings and require extensive tuning, while learning-based methods face difficulties generalizing to out-of-distribution environments. This paper introduces X-Mobility, an end-to-end generalizable navigation model that overcomes existing challenges by leveraging three key ideas. First, X-Mobility employs an auto-regressive world modeling architecture with a latent state space to capture world dynamics. Second, a diverse set of multi-head decoders enables the model to learn a rich state representation that correlates strongly with effective navigation skills. Third, by decoupling world modeling from action policy, our architecture can train effectively on a variety of data sources, both with and without expert policies: off-policy data allows the model to learn world dynamics, while on-policy data with supervisory control enables optimal action policy learning. Through extensive experiments, we demonstrate that X-Mobility not only generalizes effectively but also surpasses current state-of-the-art navigation approaches. Additionally, X-Mobility also achieves zero-shot Sim2Real transferability and shows strong potential for cross-embodiment generalization.
format Preprint
id arxiv_https___arxiv_org_abs_2410_17491
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle X-MOBILITY: End-To-End Generalizable Navigation via World Modeling
Liu, Wei
Zhao, Huihua
Li, Chenran
Biswas, Joydeep
Okal, Billy
Goyal, Pulkit
Chang, Yan
Pouya, Soha
Robotics
General-purpose navigation in challenging environments remains a significant problem in robotics, with current state-of-the-art approaches facing myriad limitations. Classical approaches struggle with cluttered settings and require extensive tuning, while learning-based methods face difficulties generalizing to out-of-distribution environments. This paper introduces X-Mobility, an end-to-end generalizable navigation model that overcomes existing challenges by leveraging three key ideas. First, X-Mobility employs an auto-regressive world modeling architecture with a latent state space to capture world dynamics. Second, a diverse set of multi-head decoders enables the model to learn a rich state representation that correlates strongly with effective navigation skills. Third, by decoupling world modeling from action policy, our architecture can train effectively on a variety of data sources, both with and without expert policies: off-policy data allows the model to learn world dynamics, while on-policy data with supervisory control enables optimal action policy learning. Through extensive experiments, we demonstrate that X-Mobility not only generalizes effectively but also surpasses current state-of-the-art navigation approaches. Additionally, X-Mobility also achieves zero-shot Sim2Real transferability and shows strong potential for cross-embodiment generalization.
title X-MOBILITY: End-To-End Generalizable Navigation via World Modeling
topic Robotics
url https://arxiv.org/abs/2410.17491