Sunday, July 17, 2016

Overselling the Automated Car

a future Mercedes-Benz?

Steven E. Shadover has written a thoughtful and provocative article  entitled "The Truth About 'Self-Driving' Cars" that appeared in the June 2016 issue of Scientific American (volume 314, pp. 52-57).

Shadover argues that the fully automated car has been oversold by both the press and that automobile industry, and that it is in reality far away from widespread adoption -- in his estimation 2075,although he concedes that it might be a bit soon, but"not by much." He does concede that cars are getting smarter -- lane avoidance capabilities, frontal crash deterrents, and more.  But to get to the final level of automated driving takes phenomenal software that will take years to develop, because of the realities of driving challenges. Snow and ice, left turns, and other challenges take the remarkable capabilities of a human brain to negotiate.  Even simple road encounters "pose huge challenges fore computers, and robotic chauffeurs remain decades away." Shadover asserts "that are pretty smart these days. Yet it is an enormous leap from sun systems [as now exist] to fully automated driving." What the automobile industry and media have generated are optimistic predictions and projections that have entered a feedback loop "that is a spiral of unrealistic expectations."

During he next 5-10 years we can look back at this article as a reference point.  Will it be cited as another pessimistic view that proved to be terribly wrong?  Or a prophetic statement that points to the limits of humans to shape their electromechanical world to suit their [well, a number of ambitious engineers and capitalists, anyway] ends?

No comments:

Post a Comment