Restricted Research - Award List, Note/Discussion Page
Fiscal Year: 2023
129 University of North Texas (142017)
Principal Investigator: He,Yanyan
Total Amount of Contract, Award, or Gift (Annual before 2011): $ 174,010
Exceeds $250,000 (Is it flagged?): No
Start and End Dates: 4/1/22 - 3/31/25
Restricted Research: YES
Academic Discipline: Mathematics
Department, Center, School, or Institute: College of Science
Title of Contract, Award, or Gift: Using uncertainty quantification and machine learning techniques to study the evolution of odor capture
Name of Granting or Contracting Agency/Entity:
U.S. Department of Defense
CFDA Link: DOD
12.431
Program Title:
none
CFDA Linked: Basic Scientific Research
Note:
Odor capture is the act of extracting chemical cues (odors) from the surrounding environment. Instead of sampling (or sniffing) with an enclosed chemosensory structure such as the sinus cavities of mammals, marine crustaceans and insects use stalks bearing arrays of chemosensory hairs. Odor capture is a complex functional system involving many parameters which interact in non-linear ways. Although detailed computational fluid dynamic (CFD) modeling has been used in the past to understand the role of hair size, arrangement, and sniffing kinematics, the limitations on computational power and speed limit the ability of traditional analysis techniques to analyze this problem. We propose the application of uncertainty quantification (UQ) and machine learning (ML) to a CFD model of odor capture to understand the role of hair-array morphology, kinematics, and fluid environment in odor capture. Additional to better describing odor-capture dynamics, understanding the evolution of these chemosensory structures requires knowledge of how variation will change the functional constraints under which biological organisms operate. These constraints are dictated by physics, environment, and phylogenetic history of an organism. Variation provides the raw materials on which natural selection acts, but the effects of variation are difficult to predict in complex, non-linear functional systems. Our combination of CFD modeling and UQ&ML techniques can map out the performance space under which these chemosensory hair arrays operate and the relative sensitivity of each parameter of odor capture to construct a global, quantitative understanding of how parameters control odor-capture performance. Furthermore, this analysis can eliminate parameters that have no influence on odor capture, extracting the root principles of odor capture and providing a more efficient way to construct bioinspired devices for chemical detection. Beyond integrating evolution and computational modeling, this work is of interest to the Army for extracting design principles that can be used for biomimetic and/or bioinspired devices for sensing chemicals in the environment. Chemical sensing of potential hazardous materials (e.g., explosives) is key to the safety and security of both civilian and military personnel worldwide. Very sensitive mechanical devices are often limited in usefulness by their size and sampling methods. Determining the best ways to miniaturize and simplify sampling routines based on design principles extracted from very successful natural systems (i.e., insects) can potentially solve these problems and lead to autonomous sampling for hazardous substances.
Discussion: No discussion notes