2026-09-19T07:10:50Zhttps://dspace.mit.edu/server/oai/requestoai:dspace.mit.edu:1721.1/1630182025-10-07T04:14:33Zcom_1721.1_7582com_1721.1_7581col_1721.1_131023
Rus, Daniela
Mishra, Kartikesh
Massachusetts Institute of Technology. Department of Electrical Engineering and Computer Science
2025-10-06T17:39:52Z
2025-10-06T17:39:52Z
2025-05
2025-06-23T14:03:03.667Z
https://hdl.handle.net/1721.1/163018
Recent vision-language navigation (VLN) approaches leverage large models, prompt engineering, and/or explicit reasoning for instruction interpretation and agent guidance. We introduce MiniNav, a minimalist framework employing frozen vision-language foundation models as patch-wise feature extractors, avoiding data and compute heavy fine-tuning and cumbersome language model reasoning. Our lightweight control policies (∼ 10⁵ trainable parameters) are trained on a compact dataset of language-based specified navigational behaviors (∼ 10² runs, ∼ 10⁴ frames per behavior). We demonstrate generalization to novel objects and scenes, including direct real-world transfer, despite training on only two objects in a single simulated environment. Through its simple and scalable design, MiniNav provides an alternative to computationally intensive pipelines for robust real-world instruction-following. Our solution can provide a reference for evaluating the effective edge of more complex and larger VLN policies.
M.Eng.
Massachusetts Institute of Technology
In Copyright - Educational Use Permitted
Copyright retained by author(s)
https://rightsstatements.org/page/InC-EDU/1.0/
Minimalist Approach to End-to-End Vision Language
Navigation with Multi-Modal Foundation Model Features
Thesis
application/pdf
Master
Master of Engineering in Electrical Engineering and Computer Science
Publication
unknown
unknown
<Publication xmlns="https://www.openaire.eu/cerif-profile/1.1/" id="43cd8589-b631-476e-b2a0-3347a40a165f">
<Type xmlns="https://www.openaire.eu/cerif-profile/vocab/COAR_Publication_Types">http://purl.org/coar/resource_type/c_1843</Type>
<Title>Minimalist Approach to End-to-End Vision Language
Navigation with Multi-Modal Foundation Model Features</Title>
<PublishedIn>
<Publication>
</Publication>
</PublishedIn>
<PublicationDate>2025-05</PublicationDate>
<Authors>
<Author>
<DisplayName>Mishra, Kartikesh</DisplayName>
<Affiliation>
<OrgUnit>
</OrgUnit>
</Affiliation>
</Author>
</Authors>
<Editors>
</Editors>
<Publishers>
<Publisher>
<DisplayName>Massachusetts Institute of Technology</DisplayName>
<OrgUnit />
</Publisher>
</Publishers>
<License>https://rightsstatements.org/page/InC-EDU/1.0/</License>
<Abstract>Recent vision-language navigation (VLN) approaches leverage large models, prompt engineering, and/or explicit reasoning for instruction interpretation and agent guidance. We introduce MiniNav, a minimalist framework employing frozen vision-language foundation models as patch-wise feature extractors, avoiding data and compute heavy fine-tuning and cumbersome language model reasoning. Our lightweight control policies (∼ 10⁵ trainable parameters) are trained on a compact dataset of language-based specified navigational behaviors (∼ 10² runs, ∼ 10⁴ frames per behavior). We demonstrate generalization to novel objects and scenes, including direct real-world transfer, despite training on only two objects in a single simulated environment. Through its simple and scalable design, MiniNav provides an alternative to computationally intensive pipelines for robust real-world instruction-following. Our solution can provide a reference for evaluating the effective edge of more complex and larger VLN policies.</Abstract>
<Access xmlns="http://purl.org/coar/access_right"
>
</Access>
</Publication>