Google’s self-driving cars are struggling to solve the puzzle of how to safely share the roads with human drivers, the New York Times reported.
According to Google, its autonomous cars have been involved in 16 crashes since 2009. Every time it was a human at fault.
Even in the one case where Google was to blame for a crash, an employee was piloting the car and not the computer.
The New York Times highlighted two cases to illustrate the problems self-driving cars faced in humans.
One of Google’s test cars was rear-ended when it slowed down to let a pedestrian cross. The safety driver then manually applied the brakes, and the car was hit from behind.
The pedestrian was fine, but the driver had to pay a visit to the hospital for mild whiplash.
Another example where Google cars have had problems is at four-way stop streets.
Autonomous cars follow the letter of the law, so it waited for the other (human-driven) cars to come to a complete stop and give it a gap to enter the intersection.
However, the other drivers kept inching forward, paralysing Google’s car in a version of the dining philosopher’s problem – with unaccounted-for human behaviour added to the mix.