Drone-Enabled Behavioral Mapping of Pedestrian-Vehicle interactions on Zebra Crossings near Schools

Abstract

As the emphasis on active mobility grows, ensuring pedestrian safety has become increasingly important. Understanding how road users behave during interactions between pedestrians and vehicles is essential for establishing effective road safety measures. AI coupled with drone technology can significantly enhance the detection and analysis of these interactions and behaviors. While traditional microsimulation methods can simulate road user behavior and potential risks, they are prone to inaccuracies since the environment is manually calibrated, and the calibration parameters may not fully represent the real-world environment. Modeling real-world trajectories and environments directly could facilitate the identification of potential risks at a micro-scale and help understand how different road users respond to soft modes and the associated infrastructure. This paper presents an innovative solution that uses drones and an AI-driven workflow to detect pedestrian-vehicle interactions and analyze vehicle behavior in high-density pedestrian areas. The system automates the detection of these interactions based on predetermined spatial and temporal conditions. Once these interactions are detected, a baseline of vehicle behavior is established by plotting the dominant speed profiles as vehicles approach zebra crossings. This baseline behavior is then used to estimate deviations for each vehicle at each movement at an unsignalized three-legged intersection. The workflow helps pinpoint behavioral anomalies’ location, cause, and temporal signature, enabling automation across extensively recorded video data. The findings highlight the potential of these disruptive technologies to assist policymakers, urban planners, and mobility experts to be aware of current traffic situations and aid in making informed decisions to enhance road safety and improve driving conditions.