The self-driving cars designed by Google Inc. had been programmed to comply with the rules of the road. Now, the Mountain View, California based firm is attempting to teach them how to drive like humans by crossing double-yellow lines, edging into intersections and cutting corners. Drivers are expected to avoid colliding with other cars; this is human nature. The problem with Google’s robots is that their digital ‘eyes’ are able to identify potential dangers and assume the worst. Therefore, they continue to tap the brakes frequently and cause other drivers to come to a halt abruptly.
Chris Urmson is heading the search engine giant’s effort to perfect the driver-less cars and in his opinion, the vehicles are a bit extra cautious than needed. In July, he spoke at a conference that they needed to add a human element to the cars. From what can be determined, the company is getting nearer to the goal of commercializing its self-driving vehicles as John Krafcik, the auto industry veteran was hired earlier this month as the chief executive of the project. Making the cars, however, remains to be a huge challenge even though they have traveled for miles on public roads and seem to work better amidst human drivers.
16 minor accidents have occurred due to the self-driving cars since 2009. 12 of them involved rear-ending the vehicles. As opposed to the national average, this is a higher accident rate, but Google asserted that the average doesn’t include the minor accidents that have occurred with the driverless cars. The company also claimed that the crashes weren’t its fault. Nonetheless, according to some, the habit of the cars to brake frequently for avoiding real yet marginal risks could be the cause. Some have said that the problem can be eliminated by Google via ‘deep learning’ techniques that aid computers in recognizing images and objects and them improving them.
The cars seem to have annoyed some drivers around the company’s headquarters. The cars move quite slowly at T intersections and they also pause on a regular basis. Furthermore, they slow down when making left turns as their software is performing calculations for a minimum safe turning distance in light of the oncoming traffic. In contrast, the risk assessment made by humans is more instinctive. The wide left turns made by the cars seem weird to passengers and cause confusion amongst other drivers.
Therefore, human turning patterns were studied by Google and its algorithms were adjusted to take more direct corners. Despite the changes that have been made like moving from a double yellow line etc., the vehicles still stop or slow down when they aren’t sure of how to proceed. Human flexibility and interpretation is now being developed by Google into the unusual cases that have arisen. The cars will sense what the others around it are doing and also take the necessary steps to adjust itself. While the company is drawing nearer to its goal, there is still time until the cars become fully operational.