Crii: FRR: intelligent ground mobile robots integrating visual and olfactory sensing for indoor odor, gas, and VOC source localization

The National Science Foundation Division of Computer and Network Systems awarded $163,896 to Louisiana TECH University on April 1, 2026, under the Computer and Information Science and Engineering program (CFDA 47.070) to develop an intelligent ground mobile robot that integrates visual and olfactory sensing for indoor odor, gas, and volatile organic compound source localization.

The project will create a semantic-level multi-modal integration framework leveraging large language models to fuse visual and olfactory information, enabling the robotic agent to generate navigation actions based on current observations and prior search history. Research will investigate methods to mitigate the simulation-to-real gap by translating simulated odor plume data into realistic representations using real-world plume data as reference. The work will provide interdisciplinary training in robotics and artificial intelligence for undergraduate and graduate students. Potential applications include safety monitoring, indoor hazard gas localization, and environmental sensing.

The award is a Project Grant assistance type with a period of performance extending from April 1, 2026, through March 31, 2028. Work is performed in Ruston, Louisiana.

Abstract

This project aims to develop an intelligent mobile ground robot that integrates visual and olfactory sensory information for indoor odor, gas, and volatile organic compound (VOC) source localization. While current robotic systems primarily rely on visual perception (e.g., cameras) to interpret the environment, olfactory sensing provides complementary contextual information that can enhance environmental understanding and source localization performance. For instance, the presence of smoke may indicate a nearby fire source, and the aroma of coffee combined with visual cues can support more accurate interpretation of environmental conditions. To address the limited integration of olfaction in robotics, this project proposes the development of an intelligent robotic agent that fuses visual and olfactory information to locate odor sources in indoor environments. This research focuses on the development of a semantic-level multi-modal integration framework that leverages large language models for decision-making. The proposed agent will generate navigation actions based on current observations and prior search history. In addition, the project will investigate methods to mitigate the SIM-to-real GAP by translating simulated odor plume data into more realistic representations using the real-world plume data as reference. The research will also provide interdisciplinary training opportunities for undergraduate and graduate students in robotics and artificial intelligence. In summary, by enabling robots to operate in contexts where vision alone is insufficient, the proposed technology has potential applications in safety monitoring, indoor hazard gas localization, and environmental sensing. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the foundation's intellectual merit and broader impacts review criteria.- subawards are not planned for this award.