Record Number of Google AV’s driving around Mountain View and Austin in September. And not a single incident
Google's self-cars have been driving around on public roads since 2009. Before July 2015, only Mountain View and other Californian public roads. In July, they've arrived in Austin (TX). (Although no incident took place there, yet).
According to September 2015 data, there are 34 Google cars driving in Mountain View's public roads and 14 in Austin's. This brings the total to 48. That's 20 more than last month and the highest number of Google vehicles on public streets yet reported.
Despite this increase, there hasn't been a single incident reported for the month of September. And because of this, the per-mile-driven incident rate dropped to 7.5 per 1,000,000 miles, down from the 7.9 per 1,000,000 miles of August.
Total Cumulative Incidents per Month
Total Incidents per 1,000,000 Mile Driven
To put this in perspective, Google's AV current incident rate is of about 7.5 per 1,000,000 miles driven. About 2.5 times the national rate for "property-damage-only crashes", as reported by the National Highway Traffic Safety Administration. Although it's true that the NHTSA statistics might overlook minor incidents.
Over Half the Accidents Occurred When a Human Driver Was Controlling the Google Car.
Only eight of the 16 total incidents occurred when the car was engaged in autonomous mode. Of the remaining, in six the car was in driving mode and in two the car was initially in auto mode, but the driver was able to take manual control of the wheel as the other vehicle approached.
This is an option that might not be possible in the future, as Google's project is that of a car "without a steering wheel or pedals".
All the incidents appear to be another car’s fault. 13 of the 16 incidents involved another car hitting the Google car from behind.
The type of events described in the report seem to indicate that self-driving technology wasn't the cause of the crash. On the contrary, they imply that a human error on the part of the driver of the other car was the likely cause of the accident.
While Google’s AV technology seems well sophisticated for public streets after more than 2 years of testing, the car itself might still need improvements. In almost all the crashes, 14 out of 16, it was the Google car that reported some damage. The other car, which likely caused the accident in the first place, was damaged only in two instances.