optics.org blog
« Eye for immigration | Main | Switch and save »
Top gear
For the last four years, Europe’s politicians and car companies have been collaborating on the Continent’s largest road safety project. In theory, the end result could be the uncrashable car, with optics and vision systems playing a crucial role.
PReVENT is an umbrella programme of nested initiatives and research projects working towards next-generation preventive and active road safety applications, encompassing a total of 56 partner organizations and a budget of more than €50 million ($79 million).
The objectives are ambitious and the acronyms are many. PReVENT aims to develop systems allowing the safe following of other cars (SASPENSE and WILLWARN), lateral support and driver monitoring (SAFELANE and LATERAL SAFE), intersection safety (INTERSAFE), and the protection of vulnerable road users and collision mitigation (COMPOSE and Use R Cams).
It’s not just Europe that’s looking to improve road safety by trusting technology rather than people. Last year we covered the Team LUX entrant in the DARPA Urban Challenge, and described the race’s daunting requirement for driverless cars to obey all traffic regulations, safely negotiate around other cars and merge into traffic.
The eventual winner was a car from Carnegie Mellon University, which employed more than a dozen lasers, cameras and radars to view the world. It averaged 14 mph, and left in its wake competitors which variously froze at intersections, turned into oncoming traffic, and two which apparently drove straight into buildings.
These sound like software issues rather than defective optics. Infrared vision systems can sense a car’s surroundings at least as well as a human can. It’s how to interpret the resulting onslaught of data that remains the problem.
Team LUX’s Jorg Kibbel was well aware of this when we spoke to him. “Our scanners have a resolution of up to 0.125°, a range of 200 m, and a field of view of about 200°. Even small objects right in front of the car can be seen. But in urban areas many complex scenarios can emerge, and the algorithms must produce the right answer. Complex crossings, lane changes, u-turns, and of course all the other robots in the race will have to be dealt with. Plus we will have to park the car.”
Until accidentally doing so in someone’s kitchen can be completely ruled out, PReVENT might have a job on its hands.
TrackBack
TrackBack URL for this entry:
http://www.iop.org/mt4/mt-tb.cgi/881