Enter or Not Enter a Building after Fire: Post-disaster Structural Safety Assessment for Immediate Rescue through UAV-acquired Images

Faculty Mentor: Dr. Mijia Yang

Large portion of structures such as buildings and bridges could be damaged in natural disasters, such as fires.  Their immediate conditions are critical for rescue activities, which could lead to precious lifesaving.  However, current technology, such as instrumentation, data collection, and diagnosis, takes time to prepare, install, and collect sensor data and could not assess the conditions of these structures on time, which causes a serious delay for lives and properties.  In this project, the PI is interested to develop a tool for post-disaster structural safety assessment through UAV images.  The technique separates in two stages.  In the first stage, static images of cracks, broken pieces, and connections of the structures will be automatically identified on the UAV-acquired images and projected to the original structures.  Through the projection, a structural analysis of the damage will be analyzed with the updated configurations and loads.  In this process, an automatic mesh generation and analysis module will be included.  In the second stage, a deformation contour of the structure will be generated through the concentrated areas or regions.  These areas will be overlaid with a safety index calculated and could be used to guide rescue workers to enter the building and rescue people.  The project will be accomplished in three tasks.  Task 1 will be developing an autopilot UAV system that could fly close to structures and avoid obstacles.  Task 2 will be researching a neural network-assisted post-disaster damage recognition based on its damage pattern through the UAV-captured images.  Task 3 will be studying structural condition assessment through the UAV-captured images based on integrated FEA analysis.  A safety index will be calculated in this task and guide the entrance of rescue workers.

Teacher and/or Community College Faculty Component: Teachers will learn (1) Auto-pilot UAVs through buildings; (2) Damage recognition through images and AI-assisted algorithms; (3) Finite element analysis of structures.  Teachers in this program will participate in the following hands-on activities: (1) Fly UAVs with graduate students; (2) Define flight routes on drone console; (3) Build models for damage recognition and structural analysis.  The research team has UAVs and the damage recognition and structural analysis software.  The graduate students will bring these devices and software and work with RET teachers hand-by-hand on how to fly UAVs and how to collect data and perform the analysis.