Abstract
Micromobility has emerged as a key solution to urban mobility challenges, offering low-carbon, space-efficient, and health-promoting alternatives to traditional transport. However, safety and infrastructure quality are still significant barriers to its widespread adoption. This paper presents the development of an integrated system that combines computer vision, GNSS geolocation, and real-time data processing to enhance micromobility safety and infrastructure monitoring. The system is composed of three specialized computer vision models: a) classification of the type of lane on which the vehicle circulates (bike lane, sidewalk, etc.), b) estimation of pedestrian density along the vehicle’s path, and c) detection of pavement and infrastructure defects, including potholes and cracks. Video feeds captured from micromobility vehicles are analyzed to identify safety hazards and infrastructure issues. Detected defects are geolocated and mapped, providing a visual interface for operators and policymakers. The system also includes a dashboard for real-time monitoring, allowing authorities to make evidence-based decisions, dynamically manage safety concerns, and optimize public space usage. By integrating data on vehicle-pedestrian and vehicle-vehicle interactions, infrastructure conditions, and traffic density, the platform supports holistic governance of shared and active mobility. This approach facilitates regulation enforcement, user awareness and promotes proactive infrastructure maintenance. Overall, the proposed solution shows how advanced sensing technologies and data-driven tools can foster safer, more efficient, and sustainable urban mobility systems, while providing tools for both operators and public authorities.
