Municipal Vehicles as Sensor Platforms to Monitor Roadway Traffic

Abstract

Urban traffic monitoring is important for applications ranging from real-time traffic-responsive signal control to long-term estimations for planning and infrastructure health. Current urban roadway traffic monitoring suffers from two problems: generally the monitoring locations are at discrete locations that are spatially sparse, and in many cases the collection lasts only a few days during a multiyear interval. As a result, most of the network is unmonitored. To supplement conventional traffic monitoring, this study explored the possibility of using municipal or public vehicles as moving sensor platforms to monitor the surrounding roadway traffic, thereby extending data collection to a large portion of the currently unmonitored network. The study used many hours of empirical sensor data collected from a prototype instrumented vehicle to identify and count passing vehicles, measure their speed, and classify the target vehicles according to measured shape and size. The instrumented vehicle is equipped with a differential global positioning system for localization and side-view, vertically scanning planar lidar sensors for perception. This paper presents the process of extracting vehicles from the lidar data, validates the results with several experiments, and ultimately demonstrates that such a system could be used to collect meaningful traffic measurements.