As the value of walking increases in the context of carbon neutrality, the shift from car-centric transportation policies to pedestrian-focused ones has gained momentum. Walking is a fundamental transportation mode that precedes the use of other transport modes, and its significance is being increasingly recognized in response to the need for sustainable, eco-friendly mobility. Policies have been developed to encourage the conversion of short-distance trips from internal combustion vehicles to walking. In Seoul, the 1st and 2nd Pedestrian Safety and Convenience Enhancement Plans have laid the groundwork for expanding pedestrian spaces and refining related laws and regulations, while the 3rd Plan focuses on enhancing the quality of the walking environment.
Despite these efforts, pedestrian safety continues to be a major concern, with a significant number of pedestrian fatalities still occurring. In 2022, pedestrian traffic fatalities dropped below 1,000 for the first time since 2009, but pedestrians still account for 35% of all traffic deaths, which is 1.9 times higher than the OECD average. Given these statistics, it is essential to effectively manage safety risks in pedestrian environments.
The need for data-driven pedestrian safety policies targeting vulnerable pedestrians, such as the elderly and children, is more pressing than ever. With the ongoing challenges posed by low birth rates and an aging population, innovative solutions are required to protect these groups. Recent advancements in scientific technologies, such as LiDAR for 3D pedestrian path data creation and digital twin technology for simulating pedestrian environments, provide new opportunities for pedestrian safety management. These technologies enable more accurate analysis and risk management, which, combined with legal and institutional reforms, make it an optimal time for implementing data-driven pedestrian safety strategies.
This study proposes a scientific framework for pedestrian safety policies targeting vulnerable groups, with a focus on pedestrian path analysis and diagnosis. The proposed framework consists of three key steps: (1) Data collection through structured and unstructured data, utilizing multiple sensors on mobile platforms, including smartphones commonly used in everyday life. (2) GeoAI-based pedestrian space analysis to detect sidewalk slope, surface material, damage, and obstacles while assessing effective pedestrian width. (3) Automatic result generation using generative AI to report findings and propose improvements based on safety guidelines.
The study also explores the feasibility of using smartphones for pedestrian environment surveys, with sensors such as accelerometers, gyroscopes, GPS, and cameras. The results show that smartphone-derived data can effectively estimate sidewalk slope and analyze surface material conditions by detecting variations in sensor data, offering a practical, crowd-sourced method for continuous data collection.
By leveraging advanced AI technologies like GeoAI and generative AI, this study demonstrates the potential to analyze and report on pedestrian environments, contributing to safer and more efficient urban mobility. The study utilizes AI Hub's datasets to analyze video footage, identifying surface damage and obstacles, and automates the generation of diagnostic reports based on standard guidelines.
Finally, this research provides valuable foundational data for Seoul’s data-driven pedestrian safety policies, contributing to the ongoing efforts in the 3rd Pedestrian Safety and Convenience Enhancement Plan. By integrating crowd-sourced data collection using smartphones, the study demonstrates the feasibility of scientifically advancing pedestrian safety policies while making data collection sustainable and cost-effective.