Objective: To identify factors that contribute to near-miss collisions between pedestrians and personal transportation devices (PTDs) in a university campus using a novel data collection method, unmanned aerial vehicle (UAV). Participants: A total of 3,349 pedestrians and 173 PTD riders were detected through UAV observations. Methods: The researchers employed UAV technology to capture and geocode the interactions and behavior of pedestrians and PTD riders. Then, a multilevel logistic regression model examined factors that contribute to near-miss collisions between pedestrians and PTDs. Results: The model outputs indicate that higher speed, non-bicycle PTDs (eg, skateboard and scooter), and some preventive actions, like reducing speed, deviating, and weaving, increase the probability of a PTD rider getting involved in a near-miss collision. Conclusions: Findings can guide campus planners to redesign the streets as a safe environment for all transportation modes and implement appropriate regulations and education programs, especially for non-bicycle PTD riders.