Spectacular recalls currently dominate the headlines. But software glitches and poor material selection are only the visible part of the problem.
First, it was a simple gas pedal. A mechanical part. Nothing an electronics engineer had to worry about. The second time, it was more relevant: a software bug caused the brakes to fail under certain circumstances. A brake failure even in rare cases is close to what nuclear power plant engineers call the maximum credible accident, so this case triggered much more attention among the bit jugglers and silicon plumbers. Everyone was surprised that both problems occurred of all companies at Toyota, the car maker that was almost a synonym for reliability.
While the discussion about the reason for the flaw was far from being completed, a new message shocked the car designer community (and probably even more the users): An airbag controller ran havoc, causing malfunction in one of the systems drivers confide their live to much like the brakes. This time it was another car vendor renowned for high quality standards, Honda. The company had to recall about half a million vehicles worldwide, more than twice as many as in the case of Toyota.
Still, it seemed that only Japanese vendors were affected by the problems. Now the recall sickness hit Volkswagen. Again, a company with high quality standards. It does not make a difference that the problems occurred at Volkswagen's South American subsidiary VW do Brasil. In cars of Volkswagen's Brazilian models Novo Gol and Voyage, a wheel bearing caused problems; in extreme cases the wheel even could loosen, creating extremely dangerous situations for the inhabitants. The vendor quickly decided to recall almost 200.000 cars.
Is the increased incidence of technical problems serious safety-relevant problems, one should add just a coincidence or does it follow a hidden regularity? If one asks if the sudden cumulation of such problems might cause the developer's community to pause for a moment and reassess operation methods, he gets a shrug as an answer. "I won't say this can't happen here too", a top manager of a large tier one told us. "But we think we do everything possible to make software as reliable and deterministic as possible", he said, pointing out that automotive software today is no longer written by programmers who define themselves as a kind of artist but by tools that crank out algorithms with predictable behavior. Software reuse makes sure that in most cases proven routines are deployed in ECUs, routines that have been tested over and over again under almost all conceivable constraints. And, last but not least, standards such as Autosar almost automatically produce structured architectures with clean separation between different functions and layers.
This is what one hears who tries to find out why the products of highly regarded developers suddenly fail. "Software irregularities were a problem ten years ago", one manager said. "Today we have tools that generate the software, and we have Autosar, so software can be considered as safe".
Really? I'm not convinced. In truth, the reason for these malfunctions persists despite all engineering efforts. It is a mixture as explosive as the one that is sparked in the combustion chamber to make a car move. Its ingredients are complexity and price pressure. Complexity is not a new phenomenon but it is still growing despite all efforts to keep it at bay. And price pressure is looming as a consequence of production overcapacities and increasing global competition. Since this mixture is ubiquitous and growing in today's car design and production, such a problem can strike again at any place, any brand, in any part.