Google has been working on building self-driving cars for a while now, they actually are already testing prototypes in the streets of some cities. They have now published their latest monthly report for these vehicles and the good news is that no accidents were reported during the month, but it also includes some interesting facts about their experience with these prototypes. Let's remember that the cars are not entirely automated yet, they include manual controls in case a human has to take over in a given situation. Since the project started in 2009, 1,268,108 miles have been driven exclusively by software and 938,621 miles were driven with human assistance. Currently, 10,000 to 15,000 miles per week are driven without human intervention on the streets, and this number is expected to increase in the coming months.
Since this was October, the company reminded us that the sensors are meant to recognize several elements in order to prevent accidents, including people and children in costumes. This is important as children's movements can be quite unpredictable. They also explained that since 2012 they signed up some Google employees that volunteered to test these vehicles, but their behavior was a little strange. The volunteers were instructed to drive those cars manually and then turn the automatic feature when they reached freeways. They also knew that they had to pay attention at all times as they would be on camera.
Reports from the volunteers were quite good, as they felt less stressed and tired from driving in traffic, even those who loved to drive appreciated what this technology had done for them. On the other hand, Google realized that they acted way too confident around this technology, which is natural as they had experienced that the technology worked. This exhibits a problem with semi-automatic cars, as studies suggest that humans can take between 5 to 8 seconds to regain control from this type of cars and it can take up to 17 seconds before people react to the alerts and recover control of the vehicle, which is simply too long when driving at high speeds and it could take even more time to understand the context. This is one of the reasons why Google is leaning towards semi-automatic cars. Of course, there are some more complications than just getting the technology working.