ToyotaTruth.com: Auto electronics bugs are there, but hard to find
Consumer Reports News: March 25, 2010 02:42 PM
There has been a lot of debate about whether electronics could be the cause of some complaints of unintended acceleration in Toyotas. While neither Toyota nor the government has been able to identify such a problem so far, presenters at an online press conference this week made the case that virtually all software has bugs, but that they can be very difficult to isolate.
The press conference was hosted by a Web site called ToyotaTruth.com and was moderated by Joan Claybrook, former administrator of the National Highway Traffic Safety Administration (NHTSA). Participating in the panel discussion were ToyotaTruth.com blog authors Dr. Antony Anderson, an electrical engineering consultant; Dr. Keith Armstrong, an electrical engineer; and Tom Murray, a trial lawyer who is currently suing Toyota. Also participating was Clarence Ditlow of the Center for Auto Safety, who is a board member of Consumers Union, the publisher of Consumer Reports. (Watch the event video, after registering.)
Auto electronics may not be fully redundant. Photo: ToyotaTruth.com
Armstrong noted that even the software used by NASA in the space shuttle, which is considered by Carnegie Mellon University to be the most robust in the world, has one undiscovered error in every 10,000 lines of code. For perspective, the electronic systems in a typical car today have 20 million lines of software code. At that rate, cars could have as many as 2,000 electronic bugs each. Therefore, the odds against finding latent bugs can be extremely high.
Using raw numbers from a peak year for unintended acceleration complaints, Armstrong noted that the NHTSA received 3,000 such complaints from 1989 to 1990. Using those statistics, he calculated it would take about 3,120,000 hours or 200 million miles of driving to replicate a problem, given its rarity. Even if you had 36 cars, no matter the brand, they would have to be driven around the clock for 10 years to replicate one instance of unintended acceleration. And those are just odds. It could take longer to replicate the exact circumstances that cause such a fault.
For reference, Consumer Reports tests about 80 new cars a year, and puts an average of around 7,000 miles on each of them, for about 560,000 miles a year. At that rate, it would take us 357 years to experience one unintended acceleration episode. (Learn about how we test cars.)
Since many complaints are generated by driver error or mechanical problems such as floor-mat entrapment, trying to isolate an example of unintended acceleration caused by a electronic malfunction would likely take much longer. Exposing vehicles to concentrated laboratory testing, similar to the accelerated durability testing of vehicles and their components, would likely speed the discovery of problems. However, the numbers still show that finding an intermittent fault is an immense and potentially near-impossible task, as the cause may be undetectable after the incident.
According to the presenters, the fact that an electronic problem hasn't been found doesn't mean it's not there. They said safety experts in other industries engineer to eliminate problems, even with only circumstantial evidence of their existence.
An engineering view of automotive electronics The presenters made the point that the aircraft and railroad industries work with the Institute for Electronics and Electrical Engineers (IEEE) to develop quality standards for safety-related electronic systems. And that among the things they design for is independent backup circuits that use different technologies, so if one fails the other won't.
In contrast, they said, automakers consider their own electronic systems to be proprietary. Since they don't share data, the presenters said, their systems are less robust. They encouraged automakers to adopt quality standards similar to those used elsewhere in the transportation industry.
Specifically, they noted that two circuits work in parallel in Toyota's electronic throttle systems, but they use the same technology. In fact, Anderson showed that the two independent circuits are on the same board. So if the same problem is introduced at exactly the same time to both circuits, such as moisture or electromagnetic interference, it could conceivably cause a problem in both circuits that the system wouldn't detect. According to Armstrong, that's what Professor David Gilbert of Southern Illinois University, presented to Congress last month. "What Dr. Gilbert did was absolutely standard" in investigating such a fault, Armstrong said.
If such a duplicate fault occurred, Armstrong added, it also wouldn't be recorded by Toyota's Event Data Recorder, which relies on the same systems that could be malfunctioning. He noted the same could be true for brake override systems. (Read: " Black Box 101: The basics of Event Data Recorders.")
In general, Armstrong was not pointing fingers specifically at Toyota, but rather at the entire auto industry. Over the last 10 years, he said, Ford has more complaints related to unintended acceleration than Toyota does. And, as reported earlier, Dr. Gilbert's disputed test protocol overrides the electronic throttle control in other brands, as well as Toyota.
Toyota responded in a press release, saying the ToyotaTruth event was funded by a trial lawyer who had sued automakers for unintended acceleration. The company repeated its assertions that defects in its electronic throttle control system couldn't cause unintended acceleration. "Toyota engineers have comprehensively tested our ETCS under both normal and abnormal conditions including electromagnetic interference, and we have never found a single case of unintended acceleration due to a defect in the system," read the statement. (Watch a Toyota video detailing how they test cars for electronic interference.)
Be sure to follow Consumer Reports Cars blog (RSS) and Twitter (@CRcars) to keep up with the latest information and advice, also see our unintended acceleration guide.
Build & Buy Car Buying Service
Get Ratings on the go and compare 
