[LINK] itN: 'Third fatal Tesla Autopilot crash ...'

David dlochrin at key.net.au
Tue May 21 13:36:15 AEST 2019


On Tuesday, 21 May 2019 10:42:52 AEST Karl Auer wrote:

> You're assuming that monitoring the vehicle will be stressful. I don't think you know that. I also think that people are all different - I know some people who find driving very stressful.

Most of the stress of driving arises from having to closely monitor & react to what's happening out there.  Monitoring "driverless" vehicle operation introduces another complete unknown: what is the computer doing?  Therefore I suggest automated-vehicle technology currently offers no nett benefit _if_ everything is done according to the book.

Now there's no dispute that while the current technology may be good enough for most situations it's not reliable enough to dispense with a human driver altogether.  But that requirement is simply unrealistic because a human driver's attention will inevitably wander, resulting in a delay or completely inadequate response to an emergency.

The claimed benefits probably arise because the human drivers are usually not paying much attention in driverless mode.


On Tuesday, 21 May 2019 11:33:51 AEST Bernard Robertson-Dunn wrote:

> Autodriving/full self-driving cars need to be tested against exceptions, not the norm.  AFAIK that has never happened and may only happen over time in use, not the lab.

I agree.

The often-quoted statistics on the known frequency of Tesla crashes almost certainly don't reflect the inherent capability of the current technology because of the requirement for a human driver, even if an imperfect one.  We also don't know (?) what proportion of the time they're operating in driverless mode, and how often crashes are avoided by skilful human drivers in nearby vehicles.

David L.




More information about the Link mailing list