Nader: Are Glitch-Free Autonomous Vehicles Possible?

January 18, 2017

Terry Costlow

ZF CEO Dr. Stefan Sommer (left) and Nvidia's automotive VP Rob Csonger unveiled a new AI-based module powered by an Nvidia processor during CES2017. (Terry Costlow photo)

Artificial intelligence (AI) is emerging as a mainstay of automated driving technologies, sparking a steady cadence of partnerships and product announcements. But the increased implementation of AI is also driving consumer advocates including Ralph Nader to question whether its complexities will cause safety issues.

A growing number of industry suppliers feel that some form of computer learning, including deep learning and machine learning, will be needed for vehicles to navigate the challenges of driverless transportation. One reason is that AI provides more versatility and robustness than rule-based techniques which can’t easily account for all the oddities that occur on global roadways.

Veteran automotive safety crusader Ralph Nader questions the industry's ability to deploy 100-percent safe autonomous technologies. (WikiCommons image)

The CES 2017 conference highlighted broad industry-wide interest in AI. ZF and Audi both announced partnerships with Nvidia, a graphic processing unit provider that’s long focused on deep learning. Bosch, Continental and Visteon all discussed their development solutions. Toyota and Honda demonstrated AI concept vehicles.

But while AI is becoming a hot topic in automotive development circles, skepticism remains. Nader, the veteran automotive safety advocate and industry critic, said during CES that it will be quite difficult for automakers to determine whether AI-based systems can operate without dangerous glitches. He noted that defects in Takata airbags, GM ignition switches and the causes of Toyota’s sudden-deceleration problems were all simple technical technologies to debug relative to the many nuances of AI and related autonomous software.

“The maximum possible simplicity is the genius of engineering,” Nader said. “If companies can’t produce comparatively simple systems without shipping defective products, how can we expect someone to find problems with complex autonomous vehicles?”

Proponents counter that AI will provide enhanced safety for autonomous vehicles, far outweighing any glitches that occur. That’s similar to the benefits that accrued with air bags and antilock braking systems, even though there have been defects and recalls in those technology arenas.

The emergence of AI will likely follow a growing trend towards partnerships between technology suppliers that understand emergent technologies and companies that have experience creating systems that work in the complex automotive environment.

“We learn from each other,” said ZF CEO Dr. Stefan Sommer. “The auto industry needs to learn from the computing industry, about software engineering, new processor, AI and adding speed. These industries need to learn that the automotive industry is different. We each want to take the best of both worlds.”