Tuesday, December 17, 2013

Autonomous Vehicles: Human Nature Meets AI



TRENDS eMAGAZINE

The vision of a world in which vehicles drive themselves, smoothly transporting their passengers at a steady speed while maintaining a safe distance from all the other vehicles on the road, has motivated companies like Google, GM, Toyota, Nissan, Audi, Ford, Volvo, Daimler, and BMW to pour millions into efforts to develop driverless cars.
To navigate roads autonomously and safely, the vehicles typically include GPS, cameras, lasers, and radar to identify the car’s position on the road, the location of obstacles and pedestrians, and its distance from other vehicles, as well as processors to combine all the data and enable the car to react in real time.
According to reports from various manufacturers, some form of advanced automation will be available within a decade.  In the beginning, that may be limited to automated systems that take control only on highways, and to cars that park themselves in garages.  But at some point beyond that, the vision of completely autonomous traffic will become a reality.1

There are several major benefits of automated vehicles.  The list includes:

·                     Fewer accidents:  According to the World Health Organization, more than 1.2 million people throughout the world are killed in car crashes every year.  Another 50 million people are injured in accidents.  According to The National Highway Traffic Safety Administration, more than 90 percent of road crashes involve human error.  Once humans are no longer responsible for driving the car, accidents are expected to drop by an equal percentage.2
·                     Lower costs:  According to the American Automobile Association, accidents cause $300 billion in losses and 33,000 deaths on the road in the United States each year.  However, if the predictions about fewer accidents hold true, then the costs will also lower.3 
·                     Time savings:  Self-driving cars that can keep track of each other using sensors could maintain the speed limit with a minimum of distance between them.4 
·                     Greater convenience:  A car could be left at the entrance to a parking garage, where it would park itself.  Or to avoid the expense of parking at a convention center or airport, it could simply drive itself back to the owner’s driveway.5
·                     Efficiency:  Since the typical car is idle for 96 percent of the time, allowing cars to drive themselves would maximize their usage.  For example, a car could return home after dropping the owner at work, eliminating the need for a second car to take the children to school.6
·                     Faster drive times:  According to an engineering study, automation could enable four times as many cars to travel on a given amount of highway space, which would reduce the amount of time and fuel that is being consumed on traffic congestion.  The Texas Transportation Institute estimates that 5.5 billion hours and 2.9 billion gallons of fuel are used during traffic congestion. Vehicles would be automatically rerouted to less-traveled roads, and traffic congestion could be eliminated.7
·                     Higher productivity:  When autonomous cars become the norm, people will become even more productive.  Initially, a licensed driver will still need to sit behind the wheel in case the system fails.  But by within a decade, millions of drivers who once needed to remain focused on the road will be free to devote their full attention to their work as they commute to their offices or homes.8 
However, the vision of a road filled exclusively with entirely autonomous vehicles won’t be achieved anytime soon.  Before then, there will be a very rough transition period, in which licensed drivers will still need to sit behind steering wheels, ready to intervene should something go wrong.  At the same time, during this transition period there will be a mix of such semi-autonomous cars and traditional human-controlled cars sharing the same highway.
This means that a human driver of a conventional car that swerves abruptly between lanes or stops unexpectedly will demand that autonomous cars in the vicinity react much more quickly than they’ll need to respond when all traffic is perfectly coordinated.  Autonomous cars will still need human drivers to pay attention and take control of the car if necessary.  But there is a real risk that the driver of an autonomous car will depend on the car to drive itself and will fail to pay attention to the road.  If that happens, the driver won’t be able to react quickly enough to take control in an emergency.

According to Clifford Nass, co-director of Stanford University’s Center for Automotive Research, “The first generations [of autonomous cars] are going to require a driver to intervene at certain points.  It turns out that may be the most dangerous moment for autonomous vehicles.  We may have this terrible irony that when the car is driving autonomously it is much safer, but because of the inability of humans to get back in the loop it may ultimately be less safe.”9

Another potential hazard is that, over time, drivers’ skills will atrophy.  An analogy can be made to pilots; because they rely heavily on autopilot, pilots are losing some of the skills that they would have gained through experience.  In 2011, in fact, the Federal Aviation Administration suggested in a draft report that recent crashes might have been caused by pilots over-relying on automation.10
Bryan Reimer, a research scientist at MIT, has compiled a chart that shows how a driver’s performance is affected by the number of tasks he is performing.  When the driver is distracted by too many tasks, performance declines sharply, as one would expect.  However, when the driver has too few things to do, his performance also declines.  Reimer concludes that a driver who is daydreaming won’t be able to react quickly enough to take control to prevent an accident.11
To address this issue, companies will need to come up with ways to prevent drivers from losing focus.  Carmakers are likely to install software that will monitor drivers and alert them if their attention wanders. 
Another obstacle to the adoption of autonomous cars is cost.  According to Strategy+Business, the spinning LIDAR that is mounted on the roofs of most driverless cars currently costs about $70,000.12  That will price the car beyond the means of most Americans.  However, economies of scale will quickly reduce the price.  In quantities of 1 million units per year, the price should drop to $500 per car or less within less than a decade of mass production. 

Finally, objections have been raised about the appearance of automated cars, which are typically unattractive because of all of the necessary on-board equipment.  But elderly drivers who want to retain their mobility will not demand sleek styling.  And carmakers will devote more attention to aesthetics once they’ve figured out how to make the cars both functional and affordable.  Moreover, for higher-end vehicles, where price is less of an issue, styling is already a priority.  For example, BMW has replaced the LIDAR on the roof with two laser scanners and three radar sensors hidden in its front and rear bumpers. 

Based on this trend, we offer the following forecasts:
First, the number of cars on the road will actually decline when all vehicles are autonomous. 
This is because the car can drive a family member to one place, drop him or her off, and then go back and pick someone else up and bring that person to the place he or she wants to go.  In the long run, it will save families money because they will only need to purchase and maintain one car.  Furthermore, children under the age of sixteen will be able to ride in the car without an adult, which will increase their parents’ productivity.  As accidents become extremely rare, fewer vehicles will need to be replaced due to collision damage.
Second, people will use their time more productively. 
Instead of focusing solely on driving, people will be able to their use commuting time to work.  The interiors of cars could be converted into workspaces, with swiveling chairs, tables, and desks.
Third, trucking will undergo a dramatic change. 
Autonomous long-haul trucks (ALHTs) will operate 24 hours per day, unlike human drivers whose hours are legally limited.  That means companies will not only eliminate the costs of unionized truck drivers; they could also cut their truck fleets by two-thirds.  Also, each truck will be more fuel efficient.  The Center for Automotive Research predicts that driverless trucks would use 15 to 20 percent less fuel.  For all these reasons, ALHTs will quickly return the initial investment, and ultimately transportation costs will drop dramatically, resulting in lower prices for consumers.
Fourth, many jobs will be lost due to creative destruction.  
People who deliver goods for a living will no longer be able to keep their jobs once automation takes over.  Truck drivers and taxi drivers have the most to lose, because there will no longer be a need for these jobs. 
Fifth, as the vehicle industry is disrupted by autonomous cars and trucks, new competitors will dominate. 
Suppliers of parts like steering wheels will be as obsolete as the buggy whip manufacturers that were disrupted when cars replaced horse-drawn carriages.  Body shops that specialize in repairing cars damaged in accidents will see their business disappear.  The balance of power among car manufacturers will also shift as the basis of competition changes.  Just as the Big Three ruled the market when drivers coveted powerful engines, only to be usurped by the Japanese car companies when fuel economy became more important, a new set of companies may emerge at the top of the hierarchy when autonomy is a vehicle’s most important feature.
Sixth, Google may be the ultimate winner in the new automotive ecosystem. 
While Google has gotten a lot of attention for the autonomous car that has prowled 300,000 miles, It’s interesting to note that Google hasn’t designed a new car; it has simplified modified an existing Prius model with its own hardware and software.  While it’s too early to know precisely what Google is planning, it seems unlikely that the company will enter the car business.  Instead, its focus could be on developing a system that would become the standard for all car manufacturers.  In effect, it would own the operating system that controlled all of the vehicles, regardless of which company made each car, and it would capture valuable information about the owners’ activities that could provide Google with yet another profitable revenue stream.
Resource List:

1.                 MIT Technology Review, November/December 2013, “Driverless Cars Are Further Away Than You Think,” by Will Knight.  © 2013 by Massachusetts Institute of Technology.  All rights reserved.
http://www.technologyreview.com/featuredstory/520431/driverless-cars-are-further-away-than-you-think/
2.                 Ibid.
3.                 Ibid.
4.                 MIT Technology Review, May/June 2013, “Proceed with Caution Toward the Self-Driving Car,” by Will Knight.  2013 by Massachusetts Institute of Technology.  All rights reserved.
http://www.technologyreview.com/review/513531/proceed-with-caution-toward-the-self-driving-car/
5.                 FoxNews, December 16, 2011, "Google Granted Patent for Driverless Car Technology.”  2011 by FOX News Network, LLC.  All rights reserved.
http://www.foxnews.com/leisure/2011/12/16/google-granted-patent-for-driverless-car-technology/
6.                 New Geography, January 18, 2013, “Will Driverless Cars Help Us Drive Less?” by Jeff Khau.  © 2013 by New Geography.  All rights reserved.
http://www.newgeography.com/content/003406-will-driverless-cars-help-us-drive-less
7.                 MIT Technology Review, November/December 2013, “Driverless Cars Are Further Away Than You Think,” by Will Knight.  © 2013 by Massachusetts Institute of Technology.  All rights reserved.
http://www.technologyreview.com/featuredstory/520431/driverless-cars-are-further-away-than-you-think/
8.                 Forbes, July 8, 2013, “How Will Google’s Driverless Cars Change Cities? It Will Expand Them, Massively,” by Tim Worstall.  © 2013 by Forbes.com, LLC.  All rights reserved.        
http://www.forbes.com/sites/timworstall/2013/07/08/how-will-googles-driverless-cars-change-cities-it-will-expand-them-massively/
9.                 Huffington Post, September 16, 2013, “No One Understands the Scariest, Most Dangerous Part of a Self-Driving Car: Us,” by Bianca Bosker.  © 2013 by TheHuffingtonPost.com, Inc.  All rights reserved. 
http://www.huffingtonpost.com/2013/09/16/self-driving-car_n_3909069.html
10.             Daily Mail, November 18, 2013, “Is Autopilot Making Flight MORE Dangerous? By Ellie Zolfagharifard.”  © 2013 by Associated Newspapers Ltd.  All rights reserved.
http://www.dailymail.co.uk/sciencetech/article-2509316/Is-autopilot-making-flight-travel-MORE-dangerous-FAA-claims-thirds-pilots-make-mistakes-reliance-technology.html
11.             MIT Technology Review, May/June 2013, “Proceed with Caution Toward the Self-Driving Car,” by Will Knight.  2013 by Massachusetts Institute of Technology.  All rights reserved.
http://www.technologyreview.com/review/513531/proceed-with-caution-toward-the-self-driving-car/
12.             Strategy+Business, Summer 2013, “The Next Autonomous Car Is a Truck,” by Peter Conway.  © 2013 by Booz & Company, Inc.  All rights reserved.     
http://www.strategy-business.com/article/00176

No comments:

Post a Comment