Automated crossing of urban intersections with external information from connected and intelligent infrastructure

Abstract

Urban intersections are often unclear due to occlusions by buildings, vegetation, and other traffic participants. Thus, crossing an intersection or merging into a priority road at an intersection is a complex task that challenges human drivers and automated vehicles. Furthermore, traffic light signals may hamper the traffic flow leading to unnecessary energy consumption. Altogether, urban intersections are an accident hotspot with a high risk for fatalities of vulnerable road users like pedestrians or cyclists. The presented video shows results from the ICT4CART project, where two intersections near Ulm were equipped with intelligent infrastructure. One is a signalized intersection, broadcasting signal state, and prediction information to connected vehicles. They can adjust their trajectory to improve driving comfort and energy consumption based on this information. The other intersection is unsignalized and very unclear due to occlusions. However, the intelligent infrastructure consisting of ten monocular cameras detects all traffic participants, tracks their state over time, and broadcasts an environment model with this information to connected vehicles. A predictive motion planning can use this information to adapt the vehicle's trajectory to improve driving comfort, energy consumption, and safety of all traffic participants close to the vehicle.