Does computer overload threaten automotive safety? - Embedded.com

Does computer overload threaten automotive safety?

As one who reviews lots of embedded software looking for safety issues, I have been intrigued by the Toyota recall, and am anxiously awaiting the verdict on the computer controlled throttle issue.

Having been around nearly half a century, I have owned cars that ran the gamut when it comes to electronics. My first car contained NO computers and used the old tried-and-true point/condenser ignition.

My second car had electronic ignition which I was wary of but came to appreciate for its 100-percent reliability and no parts needing replacement (GM's HEI ignition control did not contain an embedded computer however ).

My third car used a carburetor coupled to an early engine control computer (ECC) to lower emissions and increase efficiency. Engine sensor data processed by the computer resulted in a pulse train output to solenoids that moved needles in or out of the carburetor's “jets to richen/lean the mixture. It was a rather crude system by today's standards that had a fail-safe mode of fully retracting the needles from the jets in case of failure.

In the 100,000-plus miles I drove that car, the computer entered fail safe only once, resulting in only slightly degraded drivability albeit worse gas mileage. That problem cleared itself after the car was shutdown and restarted. I never saw it again. My fourth car had an ECC coupled to fuel injection which made for optimum economy and easy starting in cold weather (the throttle still connected my foot to the butterfly valve in the intake system by a cable, though ).

In the 180,000 miles that I drove it, the ECC performed flawlessly, and was still going strong when I finally sold the car last year. That was a 1992 model, and it had only an ECC, no antilock brakes or other items requiring embedded computers.

With the purchase of a 2008 Mustang GT, I have finally entered the age of computer-controlled “everything” in my automobile.

I have to say that, as one who analyzes software for safety issues for a living, I do NOT like the concept.

I think we have passed the point of ridiculousness in applying embedded computers to cars.

Why use an embedded computer to control a conventional analog-style instrument panel, or a manually controlled heat and air conditioning system, or for electric windows and locks or, anything else that was available on a vehicle before the advent of computers?

I think the ultimate in idiocy in this regard is the electric windows in new Mustangs. In my car, when one opens the door, the computer lowers the window about one-quarter of an inch before allowing the door to fully open, and keeps the window there until the door is fully closed, at which point the computer raises the window to its fully closed position.

I understand why Ford does this, but we had frameless windows in the era before computers, and the windows sealed just fine. So why introduce more unneeded complexity and failure modes into the system now?

To read more, go to “The fallacy of drive-by-wire.

Dean Psiropoulos is an embedded software engineer with Honeywell Aerospace in Clearwater, Fla.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.