The Book I Read: ‘Driverless’ by Hod Lipson and Melba Kurman
For the next article in this series, we are taking a look at the new and thoroughly unresolved world of autonomous vehicles with the 2016 book Driverless by Hod Lipson and Melba Kurman. Hod Lipson is a professor of Mechanical Engineering and Data Science at Columbia University who specializes in self-aware engineering systems, and Melba Kurman is an author and public speaker who studies how artificial intelligence and similar technologies play an increasingly important and immediate role in human lives. The two previously collaborated on a 2012 book called Fabricated: the new world of 3D printing.
The book provides an overview of the state of autonomous vehicles, with a particular focus on the deep learning — we’ll get into what that means — technology that underlies it. It also covers the complex landscape of industry players and policymakers that are still grappling with how to implement these innovations.
Technology
One of the more interesting aspects of this book is the revelation that automated driving is not a new idea for engineering, especially within the context of the American interstate system. Part of the book goes into early attempts, both pre and post-WWII, where different cases were made for automated driving. This would be along the lines of a Futurama or World’s Fair presentation, followed by attempts at robotic highway control of cars that would today be considered rudimentary.
As opposed to rule-based programming, where actions are more predictable, the emergent AV technology is based on deep learning, a kind of artificial intelligence that will help these new cars classify visual objects and then hand them off to other pieces of software in an automated process. Finally, the best response in any situation would be based off of statistical reasoning. In other words, the cars will end up learning how to drive better, and become smarter with time.
As sensible as this kind of engineering might seem now, in the early days of the computer — think the 1940’s and 50’s, the idea that computers could learn to think for themselves was not taken seriously. There are many examples of this kind of foresight in the book, one of which is the work of early IBM artificial intelligence researcher Arthur Samuel, who programmed a computer to learn how to play checkers better than a human expert. Despite naysayers, his program turned out to be successful, wowing television spectators, and laying the groundwork for the more famous Deep Blue chess robot that defeated master player Gary Kasparov.
Policy
While the technology might be logically sound, the trickier part is in the widespread implementation of it. This is demonstrated through the author’s frustration in seeing how officials in the U.S. Department of Transportation approached this technology. In his view, those in government had a primitive understanding of autonomous and connected vehicles. They had rosy implications of the technology without an understanding (read: private sector and high-tech familiarity) of what the technology could offer.
Their strategy was to develop vehicle-to-infrastructure, or V2I, technologies in order to reap the benefit of these new types of cars. What is missing, though, are two conditions: fully autonomous (driverless) cars and a majority of vehicles and road infrastructure being V2I equipped. In the words of the author: “right now, we’re not close to either of those conditions.” This kind of disconnect between policymakers and available technology is becoming ever more clear since this book’s 2016 publication. It opens up several problems, including data sharing policies and safety standards. The RAND Corporation published a paper last year that serves as a guideline for this new frontier.
Impact
All told, the impact of this new technology will be widespread and unforeseen. While no one can predict what exactly the world will look like, it is certain that many jobs will be wiped out, and many new ones created in the process. All kinds of drivers, from truckers to taxi drivers and chauffeurs, are already in danger of losing their jobs, the way that many have already lost their jobs due to transportation network companies like Uber and Lyft.
AVs will also impact traffic laws, from speeding and parking to pedestrian and driver safety. The potential to save lives with this technology is a major upside, and highly touted. However, the only certain thing about the future of AVs in this regard is that the journey will be bumpy. Many different stakeholders, from lawmakers and developers to insurers and everyday citizens, will play a role in how these cars will appear in their lives. How can we value a human life in the wake of automated driving? Who or what is liable in an accident involving an autonomous vehicle? Certainly, new rules will have to be established to ensure equity and responsibility for the parties involved.
Takeaways
The major thing I took away from this book is that AV technology is sure to make great changes to our modern social landscape. It will happen gradually, but will be a very obvious addition to our lives. Whether it is for better or for worse is up to the individual, and time, to tell. In the interim, what must happen in order to allow as many good things to happen while mitigating the bad, is consistent and open dialogue between the groups responsible for making this technology a part of everyday life. This means AV developers need to be in strong communication with regulators and other bodies to ensure that desired outcomes are agreed upon and reached.
While it is hard to say what the future holds, if you are a fan of innovation and growing pains, then you have a lot to look forward to with autonomous vehicles.
