Infrastructure-Based Tracking of Road Users in Urban Intersections for Partially Available, Variable-Delay Sensor Data

摘要

Reliable and accurate tracking of road users is crucial for many applications in intelligent transport systems. Especially in urban intersections, where traffic safety, optimal traffic efficiency and the operation of autonomous vehicles are still technically challenging, the requirements on environment perception systems are very high. In this paper we address the problem of an infrastructure-based multiple object tracking system which makes use of the sensor information provided by a combination of both infrastructure and mobile sensors. Due to the characteristics of wireless V2X communication used to transfer the sensor data, the signals are unsynchronized, have variable delay times and are only partially available. Our sensor fusion approach, which is specifically designed to handle these characteristics, is validated using simulated sensor data and experimentally tested in an urban test environment.