Friday, February 12, 2016

Forget electric or hybrid or mileage. There's only one thing that matters for your next car.

What are the worst things that happen to us directly? (i.e. not things that happen to our kids)

Death and extreme suffering of course. Unfortunately, one is inevitable — and the longer that is delayed the more likely the other. Don’t blame me, I didn’t make this world.

Then there’s accidentally killing someone with a car, most likely a cyclist or pedestrian. That’s got to be in the top 3, maybe top 1. Drive a lot and bad luck or bad judgment may have its way. It feels like something we want to avoid.

Historically we had to minimize driving and/or drive with care. Now though, there’s a third option. Cars are beginning to incorporate pedestrian-collision avoidance systems like Volvo’s Mobileye or Toyota’s Pre-Collision safety [1].

Forget electric or hybrid or mileage or CO2 emissions. What I want most is something to keep me from hurting unarmored people. Until we have fully autonomous cars these warning systems are my best option — especially when they work under low vision conditions like darkness, rain, snow and fog.

So when I buy our next vehicle in 3-5 years this technology will determine what I buy. Only then will I look at electric, hybrid and the rest. Maybe now you’re thinking the same thing — imagine how you’d feel in 2019 if you go without and do hurt or kill someone [3].

We will all have this choice to make in the next few years [2]. 

- fn -

[1] This ConsumerReports review mixes systems that protect drivers with things that protect pedestrians. One I care little about, the other I care a lot about. It’s hard for a car-centric culture to think about the unarmored.

[2] Government action would help greatly - another reason to vote Dem. So would insurance company discounts

[3] Easy to imagine the advertising.

See also

Update 2/17/2016

I’ve been thinking more about what I want, which is detection of unarmored people (and animals), particularly in low visibility conditions. Assuming one can devise a system that can distinguish a bicycle from a fire bush in realtime at a useful distance the challenges are liability, user interface costs, and false positives. I think the liability barrier will require government action — in particular protection against suits for failure to warn. That probably requires a democrat for President (though Trump might do anything.)

The false positive problem rules out sound alerts; they’d become far too annoying. Who wants a warning of cyclist seen a half mile away? (Presumably the car would automatically report passing proximity violations to the authorities :-). We need a subtle but useful visual symbol. The obvious solution is a visible circle that appears on the windshield around the target; but that may require a costly windshield and expensive projection equipment. At the extreme the windshield is simply a display showing the results of enhanced vision systems — so night time appears like an overcast daytime view. That would be a hard sell.

I think one could go a long way, however, by simply placing a strip of lights around the perimeter of the windshield. Then a portion of the strip would light up to bracket targets (often multiple) above, below and to the side. That would be a subtle warning that would work for my needs.


Skater said...

Taking away the responsibility of the driver to watch what they are doing is telling them that they don't have to pay attention. The nice ads on TV show stops that any idiot could make on their own. Unless people are willing to put up with many "nuisance" stops, there is no way a computer will be able to stop for the close calls that the human brain can recognize.
I had a drunk run out in front of me (close by, from the side) and I doubt that a computer could have decided that he was not going to stay in the safe zone as I could. I was able to get on the brakes but not fast enough to avoid hitting him but I think the automated system would have hit him harder. His buddies sealed it in the police report about what happened.
The Google self driving cars have caused issues on the road because of being to tentative in their actions.

There was recently a study done that showed a decrease in accidents with teens that were driving stick shift cars. The proposed theory is that they needed to be much more engaged in the process of driving and didn't get distracted as much.

JGF said...

We'll learn a lot more in the years to come. I have a much lower opinion of the skills of the drivers I see, few are as exceptional as you.

I found it very hard to see people wearing black clothes at night. A radar system would have no such trouble. There are many UI challenges, not least balancing false alerts, but having a smart windshield put a "circle" around a cyclist isn't going to cause me problems.