Author(s): Thomas Munro, Sebastien Fregeau, Ella Hansen, Lucca Coelho, Rodrigo Armaza, Efe Sezer, Efe Kaya
Mentor(s): Masoud Malekzadeh
Institution SUU
Drones have become increasingly popular in recent years, appearing in a wide range of applications from recreational use to commercial and industrial tasks. These versatile flying devices are equipped with onboard cameras and sensors, making them ideal for capturing high-resolution imagery from various perspectives. In our research, we aim to utilize a drone as a tool for predictive maintenance, specifically targeting sidewalk infrastructure. By integrating computer vision algorithms, we plan to analyze the drone-captured footage to detect and identify high-damage areas on sidewalks. While most predictive maintenance projects involving drones are conducted outdoors, our current drone model is designed for indoor environments. As a result, we are adapting our approach to use small-scale sidewalk models and conduct controlled indoor tests. This setup allows us to simulate various types of sidewalk defects, such as cracks, uneven surfaces, and spalling, and test our computer vision techniques in a safe and manageable environment. The main focus of this project is on developing a robust image analysis system that can process real-time video data collected by the drone’s camera. We will implement computer vision methods like edge detection and image segmentation to automatically identify and classify different types of damage. These techniques will allow us to assess the severity of the defects and highlight areas that may require immediate attention. By leveraging drone technology for sidewalk inspection, we aim to provide an efficient and non-intrusive method of monitoring infrastructure. Traditional inspections are often labor-intensive, time-consuming, and may miss small defects that could worsen over time. Our real-time approach offers the potential for continuous monitoring, enabling timely maintenance decisions that could prevent costly repairs and enhance public safety. The findings from our controlled indoor experiments will serve as a proof-of-concept for the application of drone-based predictive maintenance in urban settings. If successful, this technology could be adapted for larger-scale outdoor inspections once a suitable drone model is acquired. Overall, this project showcases the integration of drone technology and computer vision in civil engineering, highlighting its potential to revolutionize the way we maintain and monitor public infrastructure.