Grantee Research Project Results
Machine Learning Calibrated Low-Cost Sensing (MLCS)
EPA Grant Number: SU839965Title: Machine Learning Calibrated Low-Cost Sensing (MLCS)
Investigators: Lary, David J , Dowdy, Chris , Kiv, Daniel , Wijerante, Lakitha
Current Investigators: Lary, David J , Simmons, Christopher W , Dowdy, Chris , Kiv, Daniel , Wijerante, Lakitha , Kim, Julia Boah , Hoang, Giakhanh Huu , Aker, Adam , Balagopal, Gokul , Waczak, John , Shofner, Berkley , Harindha Wijeratne, Lakitha Omal , Lary, Matthew , Yu, Xiaohe , Jerome, Sol , Pervez, Jaynal , Noorbakhsh, Kameron , Steele, Nicholas , Nair, Nikhil , Schroder, Jake , Hogan, Benjamin
Institution: The University of Texas at Dallas , Paul Quinn College
EPA Project Officer: Callan, Richard
Phase: I
Project Period: October 1, 2019 through September 30, 2020 (Extended to September 30, 2021)
Project Amount: $25,000
RFA: P3 Awards: A National Student Design Competition Focusing on People, Prosperity and the Planet (2019) RFA Text | Recipients Lists
Research Category: P3 Awards , P3 Challenge Area - Air Quality
Description:
In this project, it is expected that students will build, calibrate and deploy low-cost, calibrated air quality sensors for the benefit of local communities. All the data is open, made publicly available with low latency, the approach is transparent, scalable, easy to replicate, and requires minimal infrastructure costs by utilizing solar-powered sensors and a Long Range Wireless Area Network (LoRaWAN). New technologies are used to provide scientifically useful information from low-cost sensors.
Objective:
This project is designed to provide low-cost calibrated air quality sensors that can be distributed at scale and used sustainably across neighborhoods, providing scientifically useful observations for data-driven insights and facilitate data-driven decisions by communities and individual citizens. The air quality sensors are intended for promoting human health, education, transparent participation of all the stakeholders, and environmental justice.
Approach:
The approach is to carefully and individually calibrate low-cost sensors against an EPA reference using machine learning and reduce recurring infrastructure and installation costs by having the sensors solar-powered and communicating using a LoRaWAN over a 5-km radius of a LoRaWAN gateway. The gateway is also solar-powered and can use either a wired network connection from community partners or a cellular modem for network connectivity. All communication is encrypted and all data is made publicly available. Over the last few years we have demonstrated the feasibility of this approach in a recently completed National Science Foundation Geolocated Allergen Sensing Project (GASP). In the process of implementation, the collaborating graduate and undergraduate students from Physics and Computer Science will learn new skills and experience the benefit of teamwork for societal benefit.
Innovative Aspects: (1) Use machine learning for individualized and validated calibration of carefully chosen long-lifetime sensors against an EPA reference instrument. (2) Apply technology developed for NASA satellite bias detection for real-time comparison of each sensor to its neighbors to detect drift and sensor degradation.
Societal Benefit: Provide open data from an open source platform giving communities the ability to weave new data collected by "smart devices" into everyday public awareness and civic policy, with quantifiable results, which can be replicated elsewhere. The specific communities are the municipalities involved (Cities of Dallas and Plano, Texas), the citizens of the neighborhood served by the sensors, and their associated community groups,and both the high school and the historically black college served by the sensors. The data from the sensors will also be available for integration into STEM curricula.
Expected Results:
The fundamental purpose of the project is to develop an open sensor design and open-source paradigm for using machine learning for the calibration of low-cost sensors so that they can readily be deployed across communities and provide scientifically useful monitoring. The data collected will be made publicly available with low latency through an open data portal built by students in another ongoing project. Fine-grained spatial and temporal observations make it easier to identify local pollution sources.
Publications and Presentations:
Publications have been submitted on this project: View all 4 publications for this projectJournal Articles:
Journal Articles have been submitted on this project: View all 2 journal articles for this projectSupplemental Keywords:
Low-cost sensors, calibration, machine learning, solar poweredProgress and Final Reports:
The perspectives, information and conclusions conveyed in research project abstracts, progress reports, final reports, journal abstracts and journal publications convey the viewpoints of the principal investigator and may not represent the views and policies of ORD and EPA. Conclusions drawn by the principal investigators have not been reviewed by the Agency.