A few weeks ago everyone was in a tizzy over Google’s announcement that they had made some cars that could drive themselves. With the caveat that I have a growing chip on my shoulder about corporations copying academic work and getting lots of good press for “inventing” stuff a good 5-10 years after government funded grants and awards (in this case primarily DARPA) have already done so I found, the ruckus to be way overhyped as driving on well mapped roads has been old hat for a few years.
Well now a European project has blown the existing ones away: an 8,000 mile journey from Italy to China without any preprogrammed map. The scale and number of conditions faced is incredible; truly amazing considering that the first DARPA Grand Challenge had no finishers and that was just in 2004. To be fair the DARPA routes where in the desert with tough terrain instead of roads, but the obstacle detection and AI has progressed faster in six years than I expected in fifteen-twenty.
At this point the primary challenge comes down to processing power and the ability to react to terrible human driving.