A Real-Time Model for Pedestrian Flow Estimation in Urban Areas based on IoT Sensors

Understanding foot traffic in real-time at a city scale is significant from different perspectives such as safety, redesigning public spaces, traffic management, hazard management, and integration of autonomous vehicles. Nevertheless, human factors are the most challenging to grasp when addressing urban dynamics, mobility, and transport problems. Moreover, accomplishing this requires extensive data on pedestrians' movement and walkable urban networks. This paper proposes an optimization-based cost-effective methodology for estimating hourly, city-wide pedestrian flow between districts, using limited pedestrian count data. Additionally, employing a network-based approach, the simulation of generated trips obtained from the flow estimation is highly accurate. We applied our method to the city of Tartu, Estonia. The results obtained in the laboratory and real field testing confirmed that our proposed model with a low error rate could be potentially applied in large urban areas with scarce data sources. Moreover, the system can benefit from real-time monitoring applications, as demonstrated in Tartu city's real-time modal split application2.