Google is trying to make its self-driving cars more human-like

Google Inc. designed its self-driving cars to follow the rules of the road. Now it’s teaching them to drive like people, by cutting corners, edging into intersections and crossing double-yellow lines. Humans expect drivers to avoid collisions. But Google’s robots assume the worst and tap the brakes frequently as their digital “eyes” spot potential dangers, sometimes prompting other drivers to stop abruptly. The cars are “a little more cautious than they need to be,” Chris Urmson, who leads Google’s effort to develop driverless cars, told a conference in July. “We are trying to make them drive more humanistically.” Google is moving closer to commercializing self-driving cars, with the hiring earlier this month of auto-industry veteran John Krafcik as chief executive of its car project. One big remaining challenge is to make the cars, which have run more than a million miles on public roads, move more seamlessly among human drivers. Since 2009, the cars have been involved in 16 minor accidents. In 12 of those mishaps, the vehicles were rear-ended. That’s a higher accident rate than the national average, but Google says national statistics exclude many minor accidents similar to those its cars have experienced.