I received a promotional copy of this book from the publisher.
Research and development of driverless cars has reached the popular press over the past few years, but until now attempts to frame the debate have remained in the specialty press and academic journals. In Driverless: Intelligent Cars and the Road Ahead, Hod Lipson and Melba Kurman offer a valuable perspective on the technological and policy implications of autonomous vehicles.
The concept of the driverless car has been around almost as long as the automobile itself, but only in the past few years has the technology underpinning the concept advanced and evolved enough to bring it close to realization. Even so, there is enough disagreement and skepticism to slow the adoption of driverless cars.
Lipson and Kurman organize their narrative around what they call the Seven Delaying Myths that slow advances in driverless car networks:
I can't address each point in depth here, but I'll make a few notes. The second myth, that technological progress is linear, is clearly false. Elementary analyses of networks show that non-linear growth occurs as the number of interconnected members increases. Those connections drive innovation through aggressive idea sharing, competition, and cooperation. The staggering growth of internet technologies and platforms puts this myth to rest easily.
The fourth point, that driverless cars require extensive investment in infrastructure, was true under the completely impractical Electronic Highway paradigm promulgated in the late 1950s and early 1960s. Lane sensors and wires embedded in the road and sensors installed in the cars were prohibitively expensive and required far more computing power than was reasonably available at the time. By 2014, the U.S. government backed research into a paradigm called V2X, where cars exchanged data with other cars, the road, and roadside sensors. Even though the available technologies and processing power were exponentially better than what was available in the 1960s, 1970s, and 1980s, the V2X system used a top-down approach where the system, writ broadly, managed each car's behavior.
One of the authors attended a 2014 U.S. Department of Transportation conference on autonomous vehicles and was astounded to see just a single session of the multi-day event devoted to Google's self-driving cars and deep learning algorithms. Disagreeing with the DOT's top-down approach (noted by including the phrase intelligent cars in the book's subtitle), the authors believe that putting the smarts and sensors in the cars and using the highway's infrastructure as a series of checkpoints and information relays is the superior solution. I find their argument persuasive. Advances in deep learning and agent-based models let individual vehicles build their skills, which they can combine with other vehicles' experiences to develop an ever-improving ensemble model through a process the authors call fleet learning.
Driverless vehicles have started to appear on American roads, but significant objections remain. What Lipson and Kurman label as Myth #6, that driverless cars need to have a perfect driving record to be safe enough, poses two problems. The first is that it's easy for critics to move the goal posts. Whatever safety level driverless cars have attained, it's easy to use the specter of a runaway or hacked vehicle a passenger has no way to control to argue that the cars must be even safer. Second, humans are horrible drivers. According to World Health Organization figures updated in May 2014, 1.2 million people are killed in car accidents worldwide every year.
And yet, even though driverless cars offer the prospect of safer roads, the loss of privacy and autonomy weighs heavily in the balance. While Myth #3, that the public is resistant, is less true than it was, a significant proporation of Americans identify strongly with their car and see it as a way to maintain their freedom. Leaving the driving to a robot would deprive those individuals of an activity they cherish, which is an attitudinal barrier policy makers can't ignore.
Driverless is an excellent book that offers a systematic and informative narrative on the history, state of the art, and future of driverless cars. Framing the issues through their Seven Myths offers a lens into the rhetoric supporting innovation and adoption of autonomous vehicles. There is much work to do on both the technological and policy sides—Lipson and Kurman's work contributes meaningfully to that discussion.
Curtis Frye is the editor of Technology and Society Book Reviews. He is the author of more than 30 books, including Improspectives, his look at applying the principles of improv comedy to business and life. His list includes more than 20 books for Microsoft Press and O'Reilly Media; he has also created more than 50 online training courses for lynda.com. In addition to his writing, Curt is a keynote speaker and entertainer. You can find more information about him at www.curtisfrye.com and follow him as @curtisfrye on Twitter.