Google Makes Autonomous Car More Human
Google first designed its autonomous cars to drive following the exact rules of the road. However, after several human-error rear-endings and a few seemingly-unnecessary brake checks, Google has decided to start programming the automatic car to drive like people.
It’s So…Beautiful *Sniff*: What drew Mazda and Toyota together?
The best example is in corners. To begin with, Google’s self-driving car would take a turn very widely. This was, in part, due to technology restrictions at the time, as the Google car had trouble identifying pedestrians near curbs and predicting their movements.
Now, though, the vehicle is programmed to cut the corner like a human would do—staying much closer to the curb. Not only does this feel more comfortable for the car’s passengers, it helps signal what the car is going to do better than just a turn signal would.
Another situation where Google cars were changed was when the car needed to cross the center line. In normal traffic, it never would have to do so, but when presented with a car parked in the middle of the road. So, Google technicians added a method that would let the Google car cross the center line to avoid parked cars or other obstructions.
We Must Be Cautious: Here’s why Mazda CPO vehicles are safer than Used ones
These changes may seem to counter the Google car’s original intention, which was to create a higher degree of safety on the road. After all, taking a corner widely is safer than hugging the curb, and crossing the center line to pass a car can be a risky proposition.
However, in order to make autonomous vehicles a reality in a world already filled with manually-controlled vehicles, it seems that autonomous cars have to take a slight step back from driving perfection to merely human levels.
News Source: Jalopnik