You may not want self-driving cars on public roads to drive as if they’re competing in a grand prix. That’s why companies that are working on this technology have gone to great lengths to ensure that their autonomous cars are overly cautious on public roads. What if the self-driving car is on a race track, could it drive like a race car driver? That’s something that the folks at Stanford University have set out to accomplish.
Stanford University engineers have developed a neural network which enables autonomous cars to perform high-speed, low-friction maneuvers that you see race car drivers make on the track. They haven’t just done this for fits and giggles.
The researchers feel that these maneuvers will better train the systems to avoid accidents. Since 94 percent of the crashes are attributable to human error, improving the ability to make split-second decisions will help reduce the possibility of road accidents.
“Our work is motivated by safety, and we want autonomous vehicles to work in many scenarios, from normal driving on high-friction asphalt to fast, low-friction driving in ice and snow,” said Nathan Spielberg, a mechanical engineer graduate student at Stanford.
The team trained the neural network with data from 200,000 motion samples which included test drives on surfaces like ice and snow. The system was then put to the test at the Thunderhill Raceway in Sacramento Valley. The team has been encouraged by the results and now want to collect more data to find out if the system can perform just as well in any condition.