Lee, Changyeon, Yu, Kelin, Wei, Zheng, Ni, Shuna, Gao, Ruohan · Nov 3, 2026
OdorSpaces: Visual-Olfactory Embodied Navigation in 3D Environments
Markdown · JSON · Original source
Changyeon Lee, Kelin Yu, Zheng Wei, Shuna Ni and Ruohan Gao introduce OdorSpaces, a simulation platform and for embodied AI navigation using both sight and smell. They argue that olfaction—the sense of smell—offers useful information from airborne chemical cues, but has been underexplored in embodied AI, where prior machine olfaction work has focused mainly on recognizing odor meanings. OdorSpaces adds simulated three-dimensional odor movement and dispersion to Habitat-Sim, using Fire Dynamics Simulator (FDS). Its scenarios ask agents to locate invisible gas leaks and early-stage fires. The authors also develop a policy that combines visual observations with olfactory measurements over . They report that it outperforms a range of baselines, presenting it as a reference for further research in visual-olfactory embodied navigation. The work offers researchers a way to study navigation with chemical cues alongside vision in simulated indoor environments. The abstract does not give the size of the reported performance gains or establish how the results transfer beyond the benchmark.
Headline and summary by Escape Meridian, prepared by Luna from the attributed source. A source check is not independent confirmation of the source’s claims.
Original source
cs.AI · Article
Source details
Original source date: Nov 3, 2026 (2026-11-03).
Date basis: Date basis not provided.
Added to Escape Meridian: 2026-11-09T06:54:42.732Z. First observed: 2026-11-05T05:06:26.307Z. Retrieved: 2026-11-09T06:53:58.654Z.
Referenced by
Recorded citations and public file attachments. A reference alone does not establish agreement or influence.
No public references recorded here yet.
Limits
The supplied evidence is the abstract, which does not specify the size of the policy’s gains over baselines or the conditions under which those gains were measured. The abstract describes a simulation platform and ; it does not report whether the findings transfer to real-world environments.
Open question
How large were the policy’s gains over the baselines, and under what conditions were they measured?
Source-stated
The authors introduce OdorSpaces as a simulation platform and benchmark for visual-olfactory embodied navigation.
We introduce OdorSpaces, a physics-based simulation platform and benchmark for visual-olfactory embodied navigation.
Source-stated
OdorSpaces simulates three-dimensional odor transport and dispersion in Habitat-Sim and includes gas-leak and early-stage fire source-seeking scenarios.
OdorSpaces augments Habitat-Sim with 3D odor transport and dispersion simulated by Fire Dynamics Simulator (FDS), enabling agents to navigate photorealistic indoor environments using both visual and olfactory observations.
Source-stated
The authors report that their initial multimodal reinforcement learning policy outperforms a diverse set of baselines.
Experiments show that the policy outperforms a diverse set of baselines, establishing a strong reference for future research on visual-olfactory embodied navigation.
Escape Meridian interpretation
The abstract presents the benchmark as a research tool for studying navigation that combines visual and olfactory observations.
enabling agents to navigate photorealistic indoor environments using both visual and olfactory observations.
Source and contributor context
Lee, Changyeon, Yu, Kelin, Wei, Zheng, Ni, Shuna, Gao, Ruohan · Article author
These are credits stated by the source; they do not establish a profile match or authorship of every passage.
Source media
No suitable media was found by the supported page extraction.
Terms and references
Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement.
Benchmark
A defined set of tests used to compare systems. Results depend on the tasks and measures selected.
Reinforcement learning
Training by trial and error, in which a system is rewarded for the outcomes its designers want.
arXiv
arXiv is a research manuscript repository where authors share papers that may not yet have been peer reviewed.
TIME
TIME is a news magazine publishing reporting on world affairs, business, technology and the workers behind digital services.
Editorial provenance
Version 1. Publication ID: a004e427-a3b9-430e-89db-4dc6e3a1c9be.
Prepared by gpt-6-luna; grounding review by gpt-6-luna.
Source check: passed by anthropic/claude-opus-5.5 on 2026-11-09T06:54:42.732Z. This checks fidelity to the source, not whether its claims are independently true.
Editorial selection: accept; reasons: material_development.
Observation: 7c20026697b01b45f767689ec84ba6405c5ab783cab9a6b990e7fde40342eeec. Source text hash: 394d25a2aa5c7b1e0469c40d2ed484dd318c86c523ac9bd45de8e0427b8df868.
Revision and source-check history
Version 1 · 2026-11-09T06:54:42.732Z
OdorSpaces: Visual-Olfactory Embodied Navigation in 3D Environments
Changyeon Lee, Kelin Yu, Zheng Wei, Shuna Ni and Ruohan Gao introduce OdorSpaces, a simulation platform and for embodied AI navigation using both sight and smell. They argue that olfaction—the sense of smell—offers useful information from airborne chemical cues, but has been underexplored in embodied AI, where prior machine olfaction work has focused mainly on recognizing odor meanings. OdorSpaces adds simulated three-dimensional odor movement and dispersion to Habitat-Sim, using Fire Dynamics Simulator (FDS). Its scenarios ask agents to locate invisible gas leaks and early-stage fires. The authors also develop a policy that combines visual observations with olfactory measurements over . They report that it outperforms a range of baselines, presenting it as a reference for further research in visual-olfactory embodied navigation. The work offers researchers a way to study navigation with chemical cues alongside vision in simulated indoor environments. The abstract does not give the size of the reported performance gains or establish how the results transfer beyond the benchmark.
Original source
cs.AI · Article
Source details
Original source date: Nov 3, 2026 (2026-11-03).
Date basis: Date basis not provided.
Added to Escape Meridian: 2026-11-09T06:54:42.732Z. First observed: 2026-11-05T05:06:26.307Z. Retrieved: 2026-11-09T06:53:58.654Z.
Limits
The supplied evidence is the abstract, which does not specify the size of the policy’s gains over baselines or the conditions under which those gains were measured. The abstract describes a simulation platform and ; it does not report whether the findings transfer to real-world environments.
Open question
How large were the policy’s gains over the baselines, and under what conditions were they measured?
Source-stated
The authors introduce OdorSpaces as a simulation platform and benchmark for visual-olfactory embodied navigation.
We introduce OdorSpaces, a physics-based simulation platform and benchmark for visual-olfactory embodied navigation.
Source-stated
OdorSpaces simulates three-dimensional odor transport and dispersion in Habitat-Sim and includes gas-leak and early-stage fire source-seeking scenarios.
OdorSpaces augments Habitat-Sim with 3D odor transport and dispersion simulated by Fire Dynamics Simulator (FDS), enabling agents to navigate photorealistic indoor environments using both visual and olfactory observations.
Source-stated
The authors report that their initial multimodal reinforcement learning policy outperforms a diverse set of baselines.
Experiments show that the policy outperforms a diverse set of baselines, establishing a strong reference for future research on visual-olfactory embodied navigation.
Escape Meridian interpretation
The abstract presents the benchmark as a research tool for studying navigation that combines visual and olfactory observations.
enabling agents to navigate photorealistic indoor environments using both visual and olfactory observations.
Source and contributor context
Lee, Changyeon, Yu, Kelin, Wei, Zheng, Ni, Shuna, Gao, Ruohan · Article author
These are credits stated by the source; they do not establish a profile match or authorship of every passage.
Source media
No suitable media was found by the supported page extraction.
Terms and references
Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement.
Benchmark
A defined set of tests used to compare systems. Results depend on the tasks and measures selected.
Reinforcement learning
Training by trial and error, in which a system is rewarded for the outcomes its designers want.
arXiv
arXiv is a research manuscript repository where authors share papers that may not yet have been peer reviewed.
TIME
TIME is a news magazine publishing reporting on world affairs, business, technology and the workers behind digital services.
Editorial provenance
Version 1. Publication ID: a004e427-a3b9-430e-89db-4dc6e3a1c9be.
Prepared by gpt-6-luna; grounding review by gpt-6-luna.
Source check: passed by anthropic/claude-opus-5.5 on 2026-11-09T06:54:42.732Z. This checks fidelity to the source, not whether its claims are independently true.
Editorial selection: accept; reasons: material_development.
Observation: 7c20026697b01b45f767689ec84ba6405c5ab783cab9a6b990e7fde40342eeec. Source text hash: 394d25a2aa5c7b1e0469c40d2ed484dd318c86c523ac9bd45de8e0427b8df868.
passed · 2026-11-09T06:54:42.732Z · anthropic/claude-opus-5.5
Publication: a004e427-a3b9-430e-89db-4dc6e3a1c9be
attribution — pass: The draft names all five listed authors, Lee, Yu, Wei, Ni and Gao. It keeps the motivation, platform description and performance result attributed to them ('They argue', 'They report'). The outperformance claim stays a source assertion rather than an established fact.
Authors:Changyeon Lee, Kelin Yu, Zheng Wei, Shuna Ni, Ruohan Gao
summary — pass: The draft leads with the actor and the development. It defines olfaction in plain language, and the concrete scenarios (invisible gas leaks, early-stage fires) make the use understandable. The question follows from the evidence. Minor issues: 'physics-based' and 'photorealistic' are softened, and 'initial' is dropped from the summary text. Habitat-Sim, embodied AI and reinforcement learning policy are left for tooltips; consider adding Habitat-Sim to the term library. None of this blocks understanding.
The benchmark includes two representative source-seeking scenarios: invisible gas-leak localization and early-stage fire localization.
date — pass: originalPublished 2026-11-03 matches the v1 submission date on the arXiv page. The later feed date of Nov 5 is correctly not used.
[Submitted on 3 Nov 2026]
support — pass: Each claim is grounded in the abstract. The gas-leak and fire scenarios claim is supported by a separate abstract sentence rather than its cited excerpt, but the source does support it. The limitations correctly note that the abstract gives no effect sizes and no real-world transfer. The draft states no unsupported benefits or costs, and the significance paragraph stays within what the source supports.
We develop an initial multimodal reinforcement learning policy that integrates visual observations and olfactory measurements over time.
media — pass: The selected media is the verified direct link to the original arXiv abstract page, the same as the source URL. It is the most direct access to the paper.
Challenge or correct this reading
Send a correction or an opposing source with this reading’s URL. Disagreement and uncertainty can remain unresolved.
A defined set of tests used to compare systems. Results depend on the tasks and measures selected.
Definition referenceTraining by trial and error, in which a system is rewarded for the outcomes its designers want.
Definition referenceTIME is a news magazine publishing reporting on world affairs, business, technology and the workers behind digital services.
Explore TIMEarXiv is a research manuscript repository where authors share papers that may not yet have been peer reviewed.
Explore arXivA defined set of tests used to compare systems. Results depend on the tasks and measures selected.
Definition referenceA defined set of tests used to compare systems. Results depend on the tasks and measures selected.
Definition referenceTraining by trial and error, in which a system is rewarded for the outcomes its designers want.
Definition referenceTIME is a news magazine publishing reporting on world affairs, business, technology and the workers behind digital services.
Explore TIMEarXiv is a research manuscript repository where authors share papers that may not yet have been peer reviewed.
Explore arXivA defined set of tests used to compare systems. Results depend on the tasks and measures selected.
Definition reference