Google Self-Driving Car at Partial Fault for Crash

Google has confirmed that one of its self-driving cars was partially responsible for a minor crash with a bus. It's the first time the company has taken a share of the blame for a prang.

The cars operate through a range of technologies including sensors, cameras, lasers, GPS and map data. The theory is that these allow them to track the activity of other vehicles on the road and more reliably avoid crashes than cars which are subject to human driver error.

California allows companies such as Google which meet set criteria to operate self-driving or "autonomous" vehicles on public roads for testing purposes, though only on streets with a speed limit of 35 miles per hour or less. One of the key rules of the licensing is that the test cars must always have a human driver on board with the ability to immediately take control in the event of an emergency.

Google issues monthly reports on any incidents involving its cars. While they've been involved in several minor accidents, to date these have always been the result of human error by the drivers of other vehicles, such as changing lanes without warning or not paying attention when making a turn. It appears that in every case the other vehicle drove into the Google car.

The only "at fault" incident came when a test driver was using the car for a personal journey and had turned the automatic controls off. The closest the self-driving cars have come to a traffic violation so far was when a police officer pulled over one vehicle for driving too slowly, though no ticket was issued.

Collision Took Place At 2MPH

Now Google says one of its cars has driven into the side of a bus that was travelling in the same direction. Thankfully it was a very minor collision as the car was travelling at just 2 miles per hour while the bus was moving at 15mph.

Both vehicles were in the same marked lane, though it was wide enough to allow two vehicles side by side in most places. The Google car had been driving close to the curb, allowing room towards the inside of the lane. However, it slowed down when it came up to sandbags by a storm drain (meaning there was no longer two vehicle widths of space) and waited for a gap in traffic to move further inside. (Source: ca.gov)

Algorithms Make Assumptions

It seems everyone and everything involved made an assumption. The car's system calculated that a bus coming up from behind would slow down to let it pull out; the human driver of the car made the same assumption and thus didn't override the controls. However, the bus driver assumed the car would stay put until a clearer space opened up in the traffic.

Google says that "This type of misunderstanding happens between human drivers on the road every day... In this case, we clearly bear some responsibility." (Source: engadget.com)

The company says some good has come of the incident. As a result of the data gathered in this and other test drives, the automatic driving system has been tweaked to take into account that drivers of large vehicles (which are more awkward to slow or stop) are less likely to yield to traffic trying to move into its lane, or into a particular area of a lane.

What's Your Opinion?

Has Google acted responsibly by publishing details of the crash and admitting partial responsibility? Is this an acceptable result of the testing process and something that's not a big deal compared with the number of human-on-human collisions? Or is it a sign that self-driving cars aren't reliable enough for safe public use?