Slashdot Mirror
Car With A Mind Of Its Own -- Part 2
An anonymous reader writes "As a sequel to the previous Slashdot story where a car 'began accelerating to 120 mph on its own', Renault (the car manufacturer) has examined the supposed faulty car, and as many of us have suspected, no anomaly has been found (google translation). Renault will initiate a court action to discover the truth about the matter. Read more about it here (translation)."
In Holland we had a similar case, a drunk driver who killed a pedestrian in a parking garage (while driving drunk) claimed his cruise control malfunctioned and he couldn't stop the car.
Whenever people need to lie to protect themselves, they'll try to blame something they don't understand, expecting that the recipient of the story will not understand the stuff either, and thus believe them.
Ofcourse this is rather stupid, but it's just the way people are wired.
...doesn't mean there wasn't a problem. Anybody remember the Therac-25 radiation machine? After a few incidents it was examined and the first couple of times no fault was found. However, after much closer inspection they found that under just the right circumstances, if things were done in just the right order, bad things would happen. And this is a Renault we're talking about.
Of course, I still think it was user error...
I personally believe the guy was full of it, on the other hand it is also standard operating procedure to deny liability first and then investigate.
Stupid sexy Flanders.
I've heard people in these situations before saying "but I couldn't cut the engine, otherwise the servo brakes/power steering wouldn't work!". Right, so driving for miles at full throttle is far better than taking a few extra yards to stop.
I recall the major issues concerning the Airbus A320 in the late eighties. There were a number of unexplained crashes and accidents, and both the pilots and Airbus were at loggerheads because Airbus couldn't see any fault with the software and had done everything possible to make it reliable, and the pilots - including survivors of actual incidents - believed the planes had gone totally out of control.
Well, it turned out that at least one of the issues had to do with circumstances in which both pilot and plane dealt with a problem without taking into account the other's actions. As an example, if the plane tilted to the right a little too far, the plane would immediately tilt it back. The problem was so would the pilot, and the two together would over-compensate and the plane would end up dangerously tilting left. So the pilot and plane would then do exactly the same thing in the other direction. Pilot assumes plane is out of control. Plane is just trying to correct the "dumb" pilot. Result, in some cases actual disasters.
Designers have a habit of looking at designs purely in terms of a control panel hooked up to a device. However, the control panel is an interface to a device - a human being - not the end-point of the design, and designers need to be more careful to ensure that the fact a human being will be a part of the system is taken into account, at all stages of the design. Airbus, of course, can be forgiven for being one of the first to encounter a problem with this.
Whether this is relevent to the Renault incident is open to question.
You are not alone. This is not normal. None of this is normal.
What is there to find? If there was a bug in some of the software on the car computers how would they find it since it's probably in every other Renault?
"This car is exactly like all the other ones - no anomolies, nothing broken - it's fine."
Chances are the computer would have auto reset like most do and any chance of software evidence being left is gone.
This is why cars should have black boxes.
That might have been unexpected, but it's seldom lethal. Your car will drive just fine without any power steering fluid (assuming you don't burn out the pump and innards of course), as that is how all cars used to be.
If you'd accelerated a little bit, you'd have noticed it got easier to turn, as the tires have less constant contact with the ground.
Basically, the point I'm making is that if power steering fails, it's a little harder to turn. If steering-by-wire fails, the car DOES NOT turn at all, and you die. Granted, I'm sure that there would be some sort of failover in place before it was ever actually implemented, but that's the difference.
Having to put some muscle into it is a whole lot better than the wheel spinning freely with no effect.
-9mm-
It's really hard, actually. At highway speed, you're covering about 30 meters (100 feet) per second. To stop, you're looking at probably three times that distance or more.
Do you want your car to activate its emergency braking system every time it approaches a hill? Will it understand that a guardrail a hundred feet in front of you is okay because you fully intend to follow a curve?
An alarm to notify drivers that they're going to lane change into the side of the car in their blind spot could be managed, but avoiding objects on the road ahead is a fiendishly difficult problem.
Incidentally, if you're following closely enough that hard braking by the car in front will cause a collision before you can respond, then you're following too closely. Back off and/or slow down.
~Idarubicin