@zzeglen, I am with you. I would also imagine recognizing if it is a dog that's crossing a street or a toddler crossing might be difficult for computer vision... But in such cases, I suppose we should brake...
@Junko I wonder how a fully autonomous car would compute that...
First it would have to distinguish between a rabbit, a deer, or a moose. Then compute the distance and speed of the vehicle doing the tailgating, weighing that aginst the size of the deer. For the moose condition probably best to slam on the brakes no matter what.
There are some things that humans will always do better than dumb machines...
@zchrish, you raise an interesting question about falling sofa...I don't know about you, but our drivers' ed class teacher taught us NOT to brake for a rabbit -- it's better to kill the rabbit and move on, rathan a car behind you bumps into you.
I wonder how a fully autonomous car would compute that...
I think another important missing component is the notion of driver leadership. In the current situation, a figurative (or collective) leadership exists in human driven driving. In a world of fully autonomous driving, who or what sets that leadership?Imagine a sofa falling out of a moving van; how would the fully autonomous vehicle set respond to it? I just find the problem rather intractable at this point. Perhaps an easier universe to test would be the set airplanes taking off and taxi-ing. It seems a vastly simpler world to model; what do airline pilots think of fully autonomous jumbo jets?
@Bert...very good point....in both cases the pilots relied too much on the automation, even when it was faulty. And some of the cases in ACI leave you wondering how the designers were thinking - as you say redundant sensors would be basic, as would warnings as to when there was a possibly bad sensor. (This would imply 3 sensors and a voting system - which of two different sensor readings do you believe?)
But how difficult it is to design a temp sensor into the pitot tubes that, when the temperature is near or below zero, switches on a heater of sufficient power to melt any ice? I could do that with the little PICAXE chips I have been playing with - which means it's not rocket science.
There was another instance where maintenance guys had taped over a pressure sensor for some reason and forgotten to take the tape off.
As is often pointed out in Air Crash investigations, there are almost always 2 problems in any crash - one usually being a malfunction of something, by accident or omission - and the other being no or incorrect corrective action being taken, either by the pilots or by the automation.
One of Arthur C Clarke's books described humans as "Superb all-purpose fault finding devices". True, but they have to know what to look for, and on an aircraft there's not always a lot of time to do so.
NASA's Orion Flight Software Production Systems Manager Darrel G. Raines joins Planet Analog Editor Steve Taranovich and Embedded.com Editor Max Maxfield to talk about embedded flight software used in Orion Spacecraft, part of NASA's Mars mission. Live radio show and live chat. Get your questions ready.
Brought to you by