Tuesday, February 9, 2016

Self Driving Cars - the missing stats

In January I described the results of my December survey on autonomous vehicles and the next day another surprising report came out on this topic. Google - a major promoter of this technology - often mentions that their cars have been involved in a tiny number of accidents and the majority were not serious or the other driver's fault.

It turns out there is a more important metric that was recently reported - how often did the test driver have to take control when the car "requested" or the driver decided on their own to do it. Companies reported many such incidents but  there is no way to know how many avoided accidents but it is safe to assume this some of them did. It is important to remember that Google's big change in this area was to switch from cars with driver controls to ones that are completely autonomous - no steering wheel in sight. They decided it would be impossible for most ordinary drivers - not the test drivers in their current vehicles - to be sufficiently alert to respond to emergencies.

We know that airplanes can and do fly themselves much of the time but there is still a pilot in the cockpit. These pilots are highly skilled and take their jobs more seriously than the average human cruising down the freeway. In spite of these unique circumstances, there is growing concern that airplanes are so automated that even a conscientious pilot may not react properly in an emergency.

As a side note - much of the research in this area is focused on tech enhancements that improve the driving experience and safety but stop far short of autonomous operation. These changes range from self-parking technology to adaptive cruise control that alters speed based on the vehicle ahead.

This Forbes report is a good read on this topic.

No comments:

Post a Comment