Essentially Volvo's approach: "Here is an extraordinary statistic: since the Volvo XC90 went on sale in the UK in 2002 it has sold over 50,000 vehicles, yet not a single person has been killed while driving it, or as a passenger."

http://www.bbc.com/news/business-43752226

That's a curious turn of phrase, and immediately makes me ask: how many other people have been killed by Volvo XC90s? The XC90 is a big, tall, heavy car that makes me nervous as a cyclist - does it protect its occupants at the expense of other road users?

A car of any size can kill a person on the road, it’s not something you design for.

well, Europe (in general, because I don't know the specifics off the top of my head) regulations require(d) changes to hood/bumper/front-end designs to improve pedestrian survivability during impacts.

I know that they require 1" (or 2.54cm) of space between hood and engine, to create a little crush zone before people hit parts that won't deform.

That's the only thing I know about it, and my source is a review of a C7 Corvette ZR1 (which Chevy cannot sell in the EU to my knowledge - due to the fact that the engine forms part of the hood)

Unlikely, given Volvo has publicly stated (https://www.volvocars.com/en-ca/about/our-stories/vision-202...):

”Our vision is that by 2020 no one should be killed or seriously injured in a new Volvo car.”

I'll take that with a grain of salt. There's a lot less opportunity to strike a deer while going 90mph on a highway in BFE or SnapChat your way into rear-ending a Chevy Suburban in the UK than there is in the US.

> I fully agree. I wish the "autopilot" would just be used passively, e.g. auto-braking when it spots something bad ahead, or some sort of warning if it notices the driver deviating from the lane (and maybe only active avoidance if that deviation means avoiding crashing into something in the other lane).

That's what other companies with driving assistance features do. The system tries to keep you safe, but it will complain rather loudly if you release the steering wheel for longer than a few seconds, it's neither intended nor sold as a self-driving system.

In many other industries, automation or alternate data/program paths are kept "on" but in a simulation system together with actual driving. So there should be a period of 5 to 10 years where all tesla cars have the autopilot system active, but controlling a virtual car in a virtual environment. As far as autopilot is concerned it's doing stuff, but only the driver is in control of the vehicle (it's best to disconnect the autopilot system completely from any physical car control systems)

Collect data and see where the two systems (one human, one automated) differ and where they react the same, and from there extrapolate how well the system works (vs a baseline of human drivers under all conditions)

Do this for 10 years and you'll have a pretty good idea of how to improve autopilot.

I'm surprised such a study hasn't been done inside telsa itself.

The space shuttle if I remember correctly, does something similar to this with the redundant processors - all computers does all calculations simultaneously, and if there are discrepancies then either use consensus (2 out of 3) or sound the emergency alarm.

That space shuttle thing is a different idea for a different problem. They use best 3 out of 5 to protect against transient corruption primarily and errors in implementation as a second factor(I think, if they use a single implementation then this is a non factor).