Smartphone-Native Bidirectional Olfactory Multimodal Large Language Models: A Unified Framework for Smell Sensing, Semantic Reasoning, Scent Generation, and Closed-Loop Olfactory Interaction
Abstract
Human–computer interaction has primarily relied on visual, auditory, and tactile channels, while olfaction remains comparatively underrepresented in general-purpose mobile computing. Recent advances in electronic noses, smartphone-connected gas sensors, artificial intelligence-based scent creation, and wearable olfactory displays provide important foundations for integrating smell into digital systems. However, a unified architecture in which a smartphone-centered multimodal large language model coordinates olfactory input, semantic interpretation, context-dependent scent generation, and feedback-based correction remains a research direction requiring systematic investigation.
This paper proposes Smartphone-Native Bidirectional Olfactory Multimodal Large Language Models (SBOM-LLMs), a conceptual framework that integrates olfactory sensing and physical scent output with multimodal large language model reasoning. The proposed architecture consists of six components: a smartphone host, an olfactory acquisition interface, an olfactory representation encoder, a multimodal reasoning controller, a constrained scent-generation planner, and a physical olfactory output interface. The framework supports two complementary capabilities: translating environmental odor measurements into machine-interpretable semantic representations and translating multimodal intentions into controlled physical scent outputs. A closed-loop mechanism compares intended scent specifications with sensor measurements and available delivery telemetry, enabling iterative adjustment when appropriate. The paper formalizes the architecture, defines candidate optimization objectives, proposes an experimental evaluation protocol, and discusses safety, calibration, personalization, and hardware constraints. This work presents a conceptual research framework rather than a validated implementation. Its potential contribution lies in the integration and coordination of existing olfactory technologies with general-purpose multimodal reasoning, not in claiming to invent electronic noses or digital scent displays.
Related
- Smartphone-Native Bidirectional Tactile Multimodal Large Language Models: A Unified Framework for Tactile Perception, Reasoning, Generation, and Physical Feedback
- Smartphone-Native Bidirectional Gustatory Multimodal Large Language Models: A Unified Framework for Taste Sensing, Reasoning, Generation, and Physical Taste Feedback
- Smartphone-Native Bidirectional Pain-Related Multimodal Large Language Models: A Unified Framework for Pain Sensing, Contextual Reasoning, Safe Sensory Feedback, and Adaptive Interaction
- Multimodal Neurorehabilitation Including Auditory EEG Neurofeedback: A Proof of Concept Tested After Resection of a Giant Olfactory Groove Meningioma
- Beyond the Bowel: A Narrative Review Examining Bidirectional Mental Health Pathways Across Disorders of Gut-Brain Interaction and Inflammatory Bowel Disease
BibTeX
@misc{zhang2026smartphonenative,
title = {{Smartphone-Native Bidirectional Olfactory Multimodal Large Language Models: A Unified Framework for Smell Sensing, Semantic Reasoning, Scent Generation, and Closed-Loop Olfactory Interaction}},
author = {Jincheng Zhang},
howpublished = {Zenodo},
year = {2026},
month = oct,
doi = {10.5281/zenodo.23494215},
url = {https://zenodo.org/records/23494215},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - Smartphone-Native Bidirectional Olfactory Multimodal Large Language Models: A Unified Framework for Smell Sensing, Semantic Reasoning, Scent Generation, and Closed-Loop Olfactory Interaction
AU - Jincheng Zhang
PY - 2026
DA - 2026/10/20
PB - Zenodo
DO - 10.5281/zenodo.23494215
UR - https://zenodo.org/records/23494215
AB - Human–computer interaction has primarily relied on visual, auditory, and tactile channels, while olfaction remains comparatively underrepresented in general-purpose mobile computing. Recent advances in electronic noses, smartphone-connected gas sensors, artificial intelligence-based scent creation, and wearable olfactory displays provide important foundations for integrating smell into digital systems. However, a unified architecture in which a smartphone-centered multimodal large language model coordinates olfactory input, semantic interpretation, context-dependent scent generation, and feedback-based correction remains a research direction requiring systematic investigation. This paper proposes Smartphone-Native Bidirectional Olfactory Multimodal Large Language Models (SBOM-LLMs), a conceptual framework that integrates olfactory sensing and physical scent output with multimodal large language model reasoning. The proposed architecture consists of six components: a smartphone host, an olfactory acquisition interface, an olfactory representation encoder, a multimodal reasoning controller, a constrained scent-generation planner, and a physical olfactory output interface. The framework supports two complementary capabilities: translating environmental odor measurements into machine-interpretable semantic representations and translating multimodal intentions into controlled physical scent outputs. A closed-loop mechanism compares intended scent specifications with sensor measurements and available delivery telemetry, enabling iterative adjustment when appropriate. The paper formalizes the architecture, defines candidate optimization objectives, proposes an experimental evaluation protocol, and discusses safety, calibration, personalization, and hardware constraints. This work presents a conceptual research framework rather than a validated implementation. Its potential contribution lies in the integration and coordination of existing olfactory technologies with general-purpose multimodal reasoning, not in claiming to invent electronic noses or digital scent displays.
ER -