Malachi Massey defending his Honors Thesis
Author: Malachi Massey | Major: Computer Science | Semester: Fall 2024
My name is Malachi Massey, a graduating senior in my last semester of study for a Bachelors of Science in Computer Science in the College of Engineering. This past year I had the privilege to conduct research in Dr. Ngan Le’s Artificial Intelligence and Computer Vision Laboratory in the Computer Science department. Upon graduation I plan to work in utilities construction with a special focus on solar PV (photovoltaic) technology.
My research is focused on developing and expanding technology to automate the process of detecting obstructions on Solar PV panels in the field. It is quite common for solar panels to get blocked by objects in nature, such as leaves, branches, and dirt. Solar farms can be very large, so manually inspecting them can be laborious and pose a danger to workers who are required to enter electrified areas. Recently, inspecting solar panels with drone technology has grown in popularity because of its ease and safety, but using drones can lead to an increased risk of human error since inspecting a solar panel from a drone is much less precise than doing so in person. My research aims to develop datasets of obstructed solar panels captured in both visible and infrared light bands, and then use Artificial Intelligence to inspect those images to automatically identify any relevant obstructions. This could vastly improve the accuracy of drone inspections and encourage the further use of this technology to save time and potentially lives.
I knew I wanted to work with AI driven Computer Vision ever since I enrolled in honors as this was the topic that initially sparked my interest in Computer Science. Dr. John Gauch recommended Dr. Le’s laboratory for me to conduct research in due to its rigorous research resume. When meeting with Dr. Le for the first time, she was excited to hear about my experience in the solar construction industry. She recommended that I work on an ongoing project that was harnessing the Transformer deep learning architecture to conduct visual segmentation on solar panels. I was excited to take on this project because of my experience and the crossover with topics that interested me.
Throughout the course of my research, I have learned an incredible amount about the intricacies of the AI models that drive computer vision. Breaking down images into a language that an AI can understand is a very unique process and is something that can be very powerful when learned. I was often pushed to my limits learning about topics that initially felt far too complex to understand, but when I put in the extra time (often all-nighters) it would open up a new world of understanding that felt so empowering. To come out of this research with knowledge, not just on specifics of solar PV segmentation, but also on how a computer thinks when it “sees” something has completely changed how I interact with computer vision processes, such as self-driving cars or generative AI photo editing. I also faced many technology related issues. Flying a drone with seven different light sensors that takes seven different photos at a time and then converting that into usable data is not an easy process. But this helped me learn a lot about image processing workflows and the work required to prepare images for computers to process them. It was a challenging, but educational, experience.
Dr. Le was a vital part of my work, giving me guidance in very difficult situations when things were not working the way they should. I learned a lot about working under pressure, and she helped guide me through the complexities of that, whether operationally or technically. I also worked with a grad student in my research who was involved in the project before I began my research. He was an integral part of everything I learned and helped guide me along the many practical steps of my research. This dynamic was one that helped me learn and grow, as well as foster independent thought and development.
As I am now graduating, I will continue to focus on the field of solar energy. I will be working for a company that constructs large-scale solar farms and will be able to leverage my research in the growing use of drones in solar development. I am so excited to continue to use the knowledge I gained in my research as I enter industry!