Autonomous Driving & Driver Assistance Systems
Driver assistance systems and autonomous driving are among the most active research areas in current vehicle development. Self-driving have the potential to considerably increase safety on public roads and offer new possibilities for modern transportation concepts.
To realize the vision of self-driving cars, the Institute of Measurement, Control, and Microtechnology develops new methods and algorithms, which render reliable driving in complex scenarios possible. This includes new approaches for environment perception, localization, situation analysis, decision making and mission planning, trajectory planning, as well as functional safety. For demonstration and testing under realistic conditions, the institute maintains two automated experimental vehicles.
The institute has two test vehicles (AV-MRM-1 in dark grey and AV-MRM-2 in white) that are both allowed and technically equipped to drive automatically on public roads. The vehicles have two standard Linux-PCs and a real-time system for safety-related functions. Based on an individual sensor concept they can perceive their environment and act accordingly by making decisions and afterwards calculating trajectories that are being processed by the control system. All relevant software modules are developed and tested at the institute.
The sensor concept of our test vehicles is based on a 360° surround view using cameras, radar sensors and laser scanners. The different sensors have only partly overlapping field of views and are based on completely different sensing methods. The main task of the sensor fusion system is thus to combine those sensors in a way that provide an optimized view of the environment.
The software modules necessary for the operation of our automated test vehicles were developed in our group and are continually improved. The figure shows the basic architecture of these modules and the information flow.
Basic Architecture of the Function Modules
We use stereo-cameras and mono-cameras, radar sensors of different resolution in both, angle and distance, as well as different lidar sensors. Further sources of information are a highly accurate digital map with localization features and other attributes as well as a GPS system, on the basis of which the exact position of the own vehicle in the map is determined by the localization module. Using all of these input channels, the current dynamic environment model is determined by means of grid mapping, multi-object tracking and classification methods. This environment model includes an individual dynamic model for each traffic participant which comprises the current dynamic state (location, speed, etc. ), class membership (cars, pedestrians, etc.) as well as a local reference to the digital map (track allocation, etc.).
The underlying module for situation understanding then determines implications between the objects. For example, a pedestrian to a pedestrian crossing is a special situation for which an automatic car has to take care of. The situation prediction module then attempts to predict the probable temporal development of the situation in order to plan the action of the automated vehicle according to the criteria of safety and comfort. Finally, the trajectory planning then determines a secure and executable vehicle trajectory that is executed by the underlying vehicle control.
The Institute for Measurement, Control and Microtechnology has the permission to operate their experimental vehicle autonomously in public road traffic. This allows to develop and test the vehicle in a real environment.
As a test route, the employees of the institute selected both, urban sections and rural sections around the university. The area covers a wide range of challenges such as passing roundabouts and intersections, the correct behavior on crosswalks and traffic lights, as well as compliance with prescribed speed limits. The illustration on the right shows the test area with markings on the respective challenges.
The application area autonomous driving and driver assistance systems has been regularly represented in the local as well as in the national press over the past years. Further, the public service broadcasting authorities showed several TV contributions about our projects.
- 16.12.2016, Südwest Presse, Mit dem Versuchsfahrzeug durch die Neue Mitte
- 18.01.2016, Pressemitteilung Ministerium für Wissenschaft, Forschung und Kunst Baden Württemberg, Meilenstein zur Entwicklung des automatisierten Fahrens
- 08.09.2015, WDR - Quarks & Co., Wenn Autos ohne Fahrer fahren
- 07.01.2015, spektrum.de, Fahrerlos durch die Innenstadt
- 07.11.2014, ARD, Fahrerloses Fahren, auf öffentlichen Straßen unterwegs
- 15.09.2014, SWR, Füße vom Pedal, Hände vom Lenkrad
- 05.08.2014 Frankfurter Allgemeine Zeitung, Noch ist der Mensch der überlegene Fahrer
- 28.07.2014, Südwest Presse, Uni Ulm testet erstes fahrerloses Fahrzeug im Straßenverkehr
- 26.07.2014, Stuttgarter Zeitung, Das muss einfach besser gehen
- 26.07.2014, Rhein Main Presse, Geisterhand am Lenkrad
- 25.07.2014 Donau3FM / youtube, Selbstfahrendes Auto in Ulm
- 25.07.2014, Augsburger Allgemeine, Auto fährt wie von Geisterhand
- 25.07.2014, Schwäbische, Führerloses Fahren ist an Uni UIm schon Realität
- 08.2013, Uni Ulm Intern, Bald führerlose Autos auf Ulmer Straßen
- 29.07.2013, automotive-technology, Autonomes Fahren: in Ulm bald führerlose Autos unterwegs
- 15.07.2013, ingenieur.de, Vollautomatisch und ohne Fahrer in der Stadt unterwegs
- 12.07.2013, swp, Bald führerlose Autos auf Ulmer Straßen