Usually, news of an automotive-related software issue involves an error like last week’s GM recall of 1 million SUVs and pickups because of a steering defect in their electric power-steering module. GM stated that the defect can cause a momentary loss of power steering followed by its sudden return, which can lead to an accident, and already has in about 30 known cases. GM says a software update to the module available from its dealers will fix the problem.
But a software remedy can’t solve Subaru’s issue with 293 of its 2019 Ascent SUVs. All 293 of the SUVs that were built in July will be scrapped because they are missing critical spot welds.
According to Subaru’s recall notice [PDF] filed with the U.S. National Highway Transportation Safety Administration, the welding robots at the Subaru Indiana Automotive plant in Lafayette, Ind., were improperly coded, which meant the robots omitted the spot welds required on the Ascents’ B-pillar. Consumer Reportsstates that the B-pillar holds the second-row door hinges. As a result, the strength of the affected Ascents’ bodies may be reduced, increasing the possibility of passenger injuries in a crash.
Subaru indicated in the recall that “there is no physical remedy available; therefore, any vehicles found with missing welds will be destroyed.” Luckily, only nine Ascents had been sold, and those customers are going to receive new vehicles. The rest were on dealer lots or in transit.
Intriguingly, the automotive manufacturer indicated that the 293 Ascents without the spot welds were assembled between 13 and 21 July, but not all the Ascents assembled during that time are missing the welds. Subaru did not provide any details on why some Ascents had missing welds, and others did not. It found an Ascent with the missing welds during a routine sampling inspection on 21 July, which immediately led to an investigation into the cause. (That must have been a real, “What the heck?” moment.)
Problems with software causing automotive manufacturing defects are rare, but not unheard of. An oft-told tale of software failure occurred in the 1980s, when GM launched a major initiative to heavily automate its car assembly lines, using robotics to compete with Japanese carmakers. However, this effort didn’t always go according to plan, as some of GM’s vehicle painting robots at its showcase luxury car Detroit-Hamtramck assembly plant painted each other rather than the cars.
The problems didn’t end there: Robots used for installing windshields had a bad habit of smashing them instead, some automated spot welders decided to weld car doors shut, and the automatic guided vehicles system used to convey parts throughout the plant never fully worked.
GM decided that it needed to scale back its automation, and gave its strategy a rethink. Apparently, though, the fundamental lessons about the limits of automation are being harshly relearned today.
In a bit of déjà vu, Elon Musk admitted earlier this year that the “excessive automation [used] at Tesla” was a mistake; instead of speeding up the company’s Model 3 production, the automation slowed it down, he said. Like GM before it, Tesla Motors had been supremely confident that extensive automation would give the company a massive competitive advantage against its rivals.
In an April interview with CBS News concerning Tesla’s Model 3 “production hell,” Musk said that, “We had this crazy, complex network of conveyor belts…. And it was not working, so we got rid of that whole thing.”
Musk confessed that Tesla had put too much technology into its Model 3 as well, saying, “We got complacent about some of the things that we felt were our core technology…. We put too much new technology into the Model 3 all at once. This—this should have been staged.”
One might wonder whether the Model 3’s technological complexities played a part in last week’s over-the-air software update that disabled several key vehicle systems including automatic emergency braking. Model 3 owners quickly noticed and complained about the problem. Tesla blamed it on a firmware issue.
An interesting question this incident raises is: If an over-the-air update erodes safety features in a fully autonomous vehicle of the future, will passengers be expected to notice?
Robert N. Charette is a Contributing Editor to IEEE Spectrum and an acknowledged international authority on information technology and systems risk management. A self-described “risk ecologist,” he is interested in the intersections of business, political, technological, and societal risks. Charette is an award-winning author of multiple books and numerous articles on the subjects of risk management, project and program management, innovation, and entrepreneurship. A Life Senior Member of the IEEE, Charette was a recipient of the IEEE Computer Society’s Golden Core Award in 2008.