Comments on: “Interest in Toyotas Starts to Revive”

By: Merovign

Merovign — Sun, 28 Mar 2010 09:51:14 +0000

A “SEU” would also only alter the acceleration until whatever the interval the device used to poll its sensors for data had passes – IIRC 2-5 milliseconds.

So, the car would accelerate uncontrollably for 1/200th of a second, then new data would replace the bad data.

Unless they used error-correcting algorithms in memory, in which case such an error would be exceedingly unlikely.

This whole kerfluffle is nothing but axe-grinding and rent-seeking, and if there were any justice at all those who have abused the system (political and legal) would be severely punished.

For the record I dislike electronic throttles, because they’re another unneeded system to fail (usually by just not working anymore), and because they’re usually programmed to dampen throttle response, which makes already over-damped modern cars feel even more sluggish.

By: Richard Nieporent

Richard Nieporent — Fri, 26 Mar 2010 16:45:27 +0000

For reasons I am unable to disclose, I am anonymously submitting several publically available scientific papers that discuss the possibility of cosmic rays disrupting electronics at sea level, essentially flipping a bit from one to zero, or vice versa. This phenomenon is a “soft” error that is not detectable except through redundant electronic and communication systems. The scientific community refers to such occurrences as “Single Event Upsets,” or SEUs. … The somewhat random nature of these events seems to imply that SEU may be one reasonable explanation for incidents of sudden acceleration.

A random single bit error would not cause sudden unintended acceleration. The way the drive by wire accelerator control works is that continuous measurements from multiple (at least two independent) sensors are used to determine the position of the accelerator pedal. A single error in a measurement caused by sea level cosmic rays would be ignored by the system. The system basically integrates over the set of measurements that it receives. An out of range data point would simply be ignored. It would take a continuous stream of bad measurements to confuse the system not a single random error. If on the other hand sea level cosmic rays caused the computer program to crash then the car monitoring system would shut off the car. If cosmic rays were the cause of sudden unintended acceleration then all of the automobile manufacturers would be experiencing the same problem which is not the case.

By: GregS

GregS — Fri, 26 Mar 2010 15:19:30 +0000

So the logic here is that if you allege a cause for a group of incidents, the *number* of incidents proves your case, even if every specific incident can be shown to be caused by something else? That’s certainly a novel notion of proof.

By: Jay

Jay — Fri, 26 Mar 2010 07:22:04 +0000

It’s a good thing they included quotes in the article from someone that owns a firm devoted to testing for and protectibg against SEU’s, otherwise I’d be tempted to think it’s just one more group trying to cash in. As it is now, I feel fairly confident in saying that I *know* it’s someone trying to cash in.