Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe
SOE: Fachverband Physik sozio-ökonomischer Systeme
SOE 22: Traffic Dynamics, Urban and Regional Systems II
SOE 22.2: Vortrag
Freitag, 4. April 2014, 10:15–10:30, GÖR 229
Autonomous Vehicle Control through Dynamic Traffic Scenarios Based on Artificial Potential Fields — •Thomas Streubel1 and Karl Heinz Hoffmann2 — 1Adam Opel AG, Rüsselsheim, Deutschland — 2TU Chemnitz, Deutschland
Mobility is a basic need in modern societies. However, the increasing traffic volume is challenging and asks for efficient solutions. Drivers are not capable of synchronizing traffic streams due to inherent limits of our cognitive abilities. Autonomous driving can overcome this restraint with sensor systems providing environmental information more accurate and faster than any driver would be able to. Here, this information is utilized to create an environmental representation in the form of an artificial potential field. In the field of robotics, it is already used for autonomous motion control. However, the vehicle environment differs greatly especially on highways due to the high speed and vehicle dynamics.
We introduce road and object models to generate the artificial potential field, so the objects as well as the road edge is assigned a high potential while the potential on the lanes is rather low. This can be interpreted as a risk map. So, the driving task is reduced to seek for a lower potential. Thus, the adverse gradient is retrieved from the potential field and is used directly for the vehicle control. Consequently, the modeling of objects and roads determines the driving behavior. The vehicle control is realized in a simulation. In particular, the lateral and longitudinal control was combined and tested in the scenario ”approaching a slower vehicle in a highway environment”.