Volvo Animal Detection – Vision 2020

Animal Detection focuses on collisions with wild animals
Of course this collection of data is not restricted to urban environments. Out in the countryside and in more remote areas there are many serious collisions involving wild animals.

Accidents involving wild animals are a major international traffic problem. In Canada, about 40,000 such accidents leading to vehicle damage are reported every year. Sweden reported 47,000 animal collisions in 2010. Of these 7,000 were elk collisions. The conditions in Canada and Sweden are also found in Norway, Finland and Russia. In the USA, about 200 people a year are killed in impacts with wild animals, mostly with deer.

However, these official accident statistics do not reveal the whole truth. For instance, they do not include all those accidents in which a driver swerves to avoid an animal and instead collides with another vehicle or veers off the road. According to a University of Umeå study of accidents between 2003 and 2010, no less than 23 percent of fatalities occurred after drivers swerved to avoid elk in the roadway – and these figures do not show up in the official statistics of collisions with wild animals.

Volvo Car Corporation is now working on Animal Detection, a system that detects and automatically brakes for animals both in daylight and in the dark (see the graphics and video).

« The technology is a further development of our pedestrian protection system. Considerable attention has been focused on ensuring that the system works in the dark since most collisions with wild animals take place at dawn and dusk, » explains Andreas Eidehall, Technical Expert Active Safety.

Accidents with wild animals often take place at cruising speeds. The aim is to reduce the speed of impact from about 100-110 km/h to below 80 km/h. Once speed drops below 80 km/h, the car’s safety systems are effective and the risk of serious injuries is small. This requires the ability to detect the animal from a distance of about 30 metres.

Another important aspect is response time – the time lapse between object identification and system reaction.
« With advanced technology we can shorten the response time still further in order to enhance the system’s effectiveness, » says Andreas Eidehall.

The system is trained to recognise the shapes of animals and their movement patterns via a vast amount of collected data. The gathering of images of animals in motion takes place on a continuous basis. But since wild animals have in many respects mastered the art of staying out of sight, this is a complex process.

What do you think?