MADISON, Wis.—Self-driving cars, if Google and automakers have their way, would already be on the road — with passengers as guinea pigs and no safety oversight in place.
If that sounds like hyperbole, think again.
High-tech and auto companies clearly prefer to let autonomous cars learn from their mistakes by driving more miles in the real world environment. The sooner those cars hit the road under minimum state and federal safety rules, the better for the corporate bottom line.
Last week during a U.S. Senate hearing, Duke University robotics professor Mary Cummings testified that autonomous cars are “absolutely not ready for widespread deployment.”
The crux of the issue, according to Cummings, is twofold: There is a lack of “principled, evidenced-based tests and evaluations” for autonomous cars, and little leadership from federal regulators to create “a clear certification process” for self-driving cars.
To make matters worse, little information about automakers’ test plans and data is available today for experts to measure the performance of self-driving cars. Put more bluntly, today’s self-driving community is substituting demonstrations for rigorous testing, in Cummings’ opinion.
In an industry and media world accustomed to the excitement among high-tech and auto companies over self-driving cars, Cummings’ statements sound a little ornery.
Cummings, however, assures that as a scientist, she “enthusiastically supports the research, development, and testing of self-driving cars.”
In her testimony, she laid out seven “limitations” in current self-driving car technologies – enough to make the public question if the current industry’s rush to deployment of autonomous cars is warranted.
EE Times followed up on Cummings’ testimony by talking to Michael Clamann, senior scientist at Duke, who heads up the university’s driverless car research. Before joining Duke, Clamann was, since 2002, a human factors engineer, supporting government and private clients in domains that included aerospace, defense and telecommunications.
1. Lack of evidence-based tests
In her testimony, Cummings said the self-driving car community is “woefully deficient” in its testing programs, “at least in the dissemination of their test plans and data.” She is concerned about the lack of “principled, evidenced-based tests and evaluations.”
Next page: Bad weather conditions