Den här sidan är utskriven från Högskolan i Halmstads webbplats ( Texten uppdaterades senast den 2015-06-04. Besök webbplatsen om du vill vara säker på att läsa den senaste versionen.


Can an automated car see?

Automated vehicles have to navigate according to their surroundings – just like a driver navigates by using her eyes. Research from Halmstad University makes it possible to create visual maps that are continuously updated. Together with these maps, cameras on an automated car determine the vehicle’s position. 

– A camera based navigation system allows the car to recognise landmarks in the area it is driving in. The problem is that surroundings change over time; weather will vary from day to day and cars in a parking lot will move. The shift between day and night, as well as seasonal variations over time, will cause drastic changes in what the area looks like, says Peter Mühlfellner, a researcher at Halmstad University.

Peter Mühlfellner, who defended his doctoral thesis at Halmstad University on May 13, has conducted his research in collaboration with a world leading car manufacturer. The research focus has been on how to build visual maps over long time frames – over one year – and under extreme appearance changes.

– The suggested method allows a constant update of an existing map, while less important information is thrown away. The storage size and processing costs for a map can therefore remain the same, but it is still possible to react to change, says Peter Mühlfellner.

Automated cars – the future of personal transportation

The system that Peter Mühlfellner and his colleagues have developed works well. The research group has built consistent maps using data collected for more than one year, in sun and rain, day and night, winter and summer. The system was recently used as a basis for automated driving in the collaborative EU research project V-Chargeexternal link.

– I believe that automated cars are a key concept for the future of individual transportation. They can help make traffic safer and more convenient. Automated valet systems, such as the one we worked on in V-Charge, can result in decreased fuel consumption and a relief of traffic congestion in cities, says Peter Mühlfellner.

The image shows how features of a parking garage can be used for localisation of automated cars.


Peter Mühlfellner defended his doctoral thesis at Halmstad University on May 13, 2015.

Photo credit: epSos.deexternal link, opens in new window, Flickr

Updated 2015-06-04