Google has been working on making its self-driving cars kid-friendly, it has recently been revealed.
On Saturday, at the company’s headquarters in Mountain View, California, children dressed in colorful and frightening costumes were invited to celebrate Halloween.
The cheerful group of robots, superheroes and other popular characters was asked to walk near parked self-driving cars, in a demonstration which revealed just how advanced autonomous technology has become.
It was also an opportunity for the company to boost the accuracy and predictive power of its algorithms even further, by having the cars face an unfamiliar situation.
“This gives our sensors and software extra practice at recognizing children in all their unique shapes and sizes, even when they’re in odd costumes”, explained company representatives in a blog post.
The exercise was particularly important because kids’ behavior tends to be more erratic and unexpected, which makes it more difficult for drivers to avoid potentially deadly collisions.
It was revealed that innovative driverless vehicles are indeed capable of anticipating the most surprising movements, thus ensuring a heightened level of safety for passengers and pedestrians likewise.
This skill of making sense out of random or chaotic situations is just as effective even when the automobile is in the parking lot, and it was made possible thanks to machine learning and artificial intelligence features.
The programs allow the cars to intelligently react to the presence of stimuli, and identify risky situations before they actually manifest themselves. For instance, the software heightens its level of security when children are nearby, to prevent accidents caused by sudden running or darting.
The tech giant, which was recently declared the second most valuable brand in the world, is still conducting trial testing of its autonomous vehicles on the streets of Texas and California.
According to company officials, the self-driving cars are in fact much more reliable and safer than regular vehicles, because they remove the possibility of human errors.
As the National Highway Traffic Safety Administration has shown, approximately 94% of all vehicle accidents occur due to mistakes or misjudgments made by drivers or pedestrians.
Ever since 2009, Google’s robotic automobiles have covered a total of 2.2 million miles, while been manually driven or autonomously driven, the equivalent of 10,000-15,000 autonomous miles on a weekly basis.
While there have been 14 accidents during those trials carried out using 50 vehicles, they all resulted from miscalculations or blunders made by other traffic participants. For example, one individual was caught playing the trumpet while driving, while another one was found engrossed in a book.
On the other hand, experts at the University of Michigan’s Transportation Research Institute has revealed that autonomous vehicles are almost 5 times more at risk of being involved in a car crash. 9.1 collisions were reported for every 1,000,000 miles, whereas normal cars have around 1.9 accidents on average.
According to researchers, this surprisingly high collision rate may be because intelligent vehicles react differently in comparison with ordinary drivers, which makes their moves more difficult to anticipate by humans.
Image Source: Google Blog