Simulation of Dense Traffic Situations to Validate Autonomous Driving Functions

摘要

Road traffic in cities at rush hour is characterized by short distances between vehicles and frequent lane changes. With the increasing automation functions in modern vehicles, the traffic participants in these scenarios have to be precisely detected. A cut in scenario with distances below 30 cm can simulate these situations. FEV describes how these scenarios can be safely simulated on a test site with GPS-controlled foam vehicles and discusses preparation, route planning, safety concept and results. Dense urban traffic in metropolises is characterized by low speeds and frequent lane changes with little distance between them. Driver assistance systems such as emergency brake assistants or congestion pilots depend on correctly determining the distances to all road users and predicting the direction of movement, as they make their decisions based on this data. This is achieved by determining the distances as well as the relative speeds of the road users with sensors such as radar, camera, lidar and ultrasonic sensors. Different sensor variants have different detection ranges and distances, both laterally and longitudinally, depending on their design. Object data from the different sensors are fused and relevant objects are used for driving control by means of a plausibility check.