AUTOMATED VEHICLES: Driverless car system developing to master road construction sites

German researchers are working with Deep Learning technology to teach driverless-vehicle software to recognize work-zone traffic patterns

March 02, 2017
German researchers are working with Deep Learning technology to teach driverless-vehicle software to recognize work-zone traffic patterns

A new system is currently under development that seeks to improve the real-time interpretation of traffic information on construction sites by autonomous vehicles.

Driverless cars have to be able to accurately detect traffic signs, but current technologies are facing difficulties when presented with temporary traffic management as appears in construction zones.

Researchers at the Fraunhofer Institute for Intelligent Analysis and Information Systems (IAIS), based in Sankt Augustin, Germany, are working with technology that is with the goal of enabling a system to read temporary traffic management signs with high accuracy. The Deep Learning technology teaches the software in autonomous vehicles to recognize the complexities presented by new traffic patterns in work zones.

The Fraunhofer team believes that through the interplay between navigation equipment and on-board computers, it will be possible for road traffic complexities on construction sites to be correctly identified, for optimal distances to be maintained between vehicles, and for speed to be adjusted in a timely manner.

In the current model, an automotive camera is used to deliver 20 to 25 frames per second. During the journey, these images are analyzed for information relating to road signs, lane information and LED traffic announcements. The future vision sees this camera as a primary interface, making many of the vehicle sensors redundant.

The project, which is expected to continue for the next three years, falls under the AutoConstruct scheme launched in December last year by the German Federal Ministry of Economics and Energy (BMWi).

The national initiative sets out to develop systems for the real-time environment recognition of construction sites by automated vehicles.

expand_less