1. 0 Introduction When the February 2012 edition of WIRED magazine appeared in my mailbox last month, the cover immediately captured my attention. “Your Next Car Will Drive Itself! ” it claimed. “No traffic jams. No crashes. Unlimited texting. ” 1 Intrigued, I immediately turned to and consumed the article finding myself fascinated by the notion of this autonomous vehicle. Alas, between work and school, I realized there would be little opportunity to develop a deeper understanding of the concepts discussed by Tom Vanderbilt.
As if to answer my silent sigh of resignation, the first set of ISTM 6406 slides appeared in Blackboard, informing me of a term paper requirement related to Decision Support Systems (DSS) and offering Artificial Intelligence (brain of an autonomous vehicle) as a specific study area. Given this opportunity to expand my knowledge of an autonomous vehicle while relating its base decision making functions to DSS, I use this term paper to provide an overview of autonomous vehicles, describe how they relate to DSS, explore their current and potential future applications, and identify some of the challenges they face.
As a wrap up, I include some of my personal thoughts on the long term viability of the autonomous vehicle. 2. 0 Autonomous Vehicles To initiate discussion, it is useful to establish common lexicon. What exactly is an autonomous vehicle? In 2003 the McGraw-Hill Dictionary of Scientific & Technical Terms defined an autonomous vehicle as “a vehicle that is able to plan its path and to execute its plan without human intervention. (McGraw-Hill Dictionary of Scientific & Technical Terms, 6E, 2003) While this definition remains accurate, the transpiring decade brings technological advances, dedicated engineers, and live demonstrations which add a degree of functional granularity to the general definition.
PCMag. com Encyclopedia now defines an autonomous vehicle as “A passenger vehicle that drives by itself. Also known as the ‘driverless car,’ it automatically steers the vehicle by sensing the painted lines in the road or a magnetic onorail embedded in the road. ” (PCMag. com Encyclopedia, 2011) 1 2 Cover of WIRED magazine, Feb 2012 (hard copy) Author of “Let the Robot Drive”, WIRED magazine, Feb 2012 2. 1 Recent History Since the 1980s, television has teased consumers with the notion of self-aware vehicles with humor, judgment, and even personality. For example, from 1982-1986, Knight Industries Two Thousand (more affectionately known as KITT) was an artificially intelligent black Trans Am partnered with Michael Knight to fight crime in NBC’s Knight Rider.
Other media highlights include Batman’s Batmobile throughout the 1990s or the generally accepted standard transportation depicted in Will Smith’s iRobot. In parallel, but outside of Hollywood’s special effects, however, true research on exactly this type of vehicle had begun. The advent of drive-by-wire technology, “electronically controlled moving parts that actuate essential components like throttles, steering, and brakes” brought with it instantaneous response and the potential for computer control.
Chrysler introduces anti-lock brakes in the 1970s. In 1988, BMW produced a drive-by-wire throttle to enable the traction-control system to adjust engine speed and limit wheel spin. (Brown, 2010) Toyota debuts radar-based adaptive cruise control, which maintains a safe driving distance from the car ahead in 1997. Through the first decade of the new millennium, Mercedes, Infinity, Volvo, Lexus, Volkswagon and others introduce All Wheel Drive, Electronic Stability Control, Dynamic Steering Response, and other drive-by-wire technologies to the market.
While most of this technology is familiar to recent car purchasers, they remain in the “driver assist” category. In 2004, 2005, and 2007, the Defense Advanced Research Project Agency (DARPA) used a series of competitions to challenge innovators and engineers to cross the line between “driver assist” and “driverless. ” The 2004 and 2005 Grand Challenges were conducted in the desert where rough surfaces, narrow borders, and irregular maneuvers confronted vehicles entered in the contest.
Though not one driverless vehicle completed the course in 2004, five managed to finish in 2005. For the 2007 Urban Challenge, DARPA increased the course difficulty by hiring professional drivers and trading the difficult but stationary terrain for smooth “roads” and a highly dynamic environment. Vehicles that participated in the 2004 and 2005 competitions made decisions on how to position themselves on a road, manage turns and narrows, and accelerate or decelerate based on sensing of road conditions, logic processes, and algorithms to identifying the “best” execution.
Those eleven that participated in the 2007 challenge were required to make more complex and real-time decisions based on predictive analysis of the behavior of other vehicles, recognition and application of rules of the road, and management of unanticipated behavior (i. e. road blockage). (DARPA Grand Challenge, 2008) 2. 2 Artificial Intelligence and Decision Support Systems The nature of the decisions made by the autonomous vehicles participating in DARPA’s 2007 Urban Challenge leveraged both Artificial Intelligence (AI) and a Decision Support System (DSS).
One of the key characteristics of AI is a system’s ability to experience and apply the knowledge – learn from empirical data. AI is also able to “intuit” missing information based on awareness of patterns among previously experienced information or analyzing for logical connections between pieces of known information. (Breimer, 2006) Alternatively, DSS are computer systems that aid people, organizations, etc. in decision making by providing ways to visualize structured data.
DSS still rely on human intuition to actually make a given decision. Since DSS present structured data in a consumable format, DSS accelerate the human decision making process in an unstructured or semi-structured information environment. In the case of an autonomous vehicle, sensors feed structured data into a DSS. The DSS displays this information as programmed. For example, in a Google car, “I watch the action unfold on the computer monitor mounted on the passenger side of the dashboard.
It shows how the car is interpreting the world: lanes, signs, cars, speeds, distances, vectors. The rendering is nothing special – a lot of blocky wireframe that puts me in the mind of Atari’s classic Battlezone. ” (Vanderbilt, 2012) Here, AI kicks in. While programmed with the basic rules of the road, AI allows the autonomous vehicle to learn from its road experiences and extend its knowledge from rigid rule-based performance to a combination of rule and experience-based performance.
AI positions an autonomous vehicle in the safest part of the lane (which may not be the center of the lane when a large semi or bus is driving in the next lane), prevents the autonomous vehicle from driving in another vehicle’s blind spot, or “if other cars don’t reciprocate [to following the rules of the road, the autonomous vehicle] advances a bit to show… other drivers its intention. Without programming that kind of behavior… it would be impossible for the robot car to drive in the real world. ” (Vanderbilt, 2012) (Guizzo, 2011) . 0 Application Shifting from a more functional discussion of autonomous vehicles to their application in the real world, this section considers both current and potential future uses of the driverless car.
3. 1 Current Employment According to IEEE Spectrum, as of October 2011, “Google’s fleet of robotic Toyota Priuses has now logged more than 190,000 miles, driving in city traffic [including Lombard Street in San Francisco], busy highways, and mountainous roads, with only occasional human intervention. (Guizzo, 2011) Google’s Priuses have experienced 2 accidents: one when it was being driven by its human pilot and one when another vehicle was at fault, rear-ending the robot vehicle at a stop light. (Cunningham, 2011)
To demonstrate these autonomous systems, Google set up driverless golf carts on its campus in Mountain View. Employees “order” a golf cart from a web browser to transport them from one part of the campus to another part. The golf carts, following the rules of the road, drive themselves to the pick-up location, where the employee embarks and chooses whether to continue in driverless mode or shift to assume control him/herself. When its mission is complete, the golf cart drives itself to a new employee pick-up location or back to the garage to wait for its next tasking. Outside of the United States, others have begun putting driverless vehicle technology to work in practical situations. For example, at London’s Heathrow airport, autonomous vehicles, called Pod Cars, transport passengers between Terminal 5’s two business parking lots. The pods carry four passengers at a time plus luggage, operate at a maximum of 25 miles per hour, and have yet to be in an accident.
Heathrow has used these 22 pod cars to replace two diesel busses, significantly increasing energy efficiency. The pod cars move only on demand vice making constant loops independent of passenger load and operate on electricity vice diesel. (Witkin, 2011) 3. 2 Potential Future Employment Google’s campus driverless golf cart ride-share program and Heathrow’s energy saving Pod Cars are only the beginning. Truly, the potential applications for autonomous vehicles are 3 Embedded video in the “How Google’s Self-Driving Car Works” article by Erico Guizzo at IEEE Spectrum ound to change some basic societal standards that currently exist. In the face of multi-state legislation efforts to ban texting while driving, “maybe the problem is not that texting and Facebook are distracting us from driving. Maybe the problem is that driving distracts us from our digital lives. ” (Vanderbilt, 2012) 3. 2. 1 Road Use Efficiency Because autonomous vehicles are smarter and faster reacting than human drivers, roads dedicated to the former enable vehicles to travel more closely together both side-to-side and front-to-rear.
Robot cars mitigate the need to maintain greater stopping distances humans require to see, process, recognize, and react to hazards ahead of them. Lanes in roads can be narrowed, as a vehicle’s autonomy eliminates a human driver’s tendency to drift in a lane. As a result, autonomous vehicles would make “better use of the 80 percent to 90 percent of empty space on roads. ”4 3. 2. 2 Fuel Savings Minimizing inconsistencies initiated by human drivers (i. e. slowing unnecessarily or over-braking in heavy traffic situations), autonomous vehicles reduce the accordion effect that causes stop-and-go traffic on major highways throughout the country.
Less time spent idling or crawling on roads increases fuel efficiency and decreases overall consumption. Additionally, broadening Google’s campus golf cart system described in the Current Employment section above, these vehicles can extend the current ride sharing functions employed in most major cities today. Where today’s ride share programs require users to locate a vehicle and walk or otherwise get to the vehicle, users would be able to reserve or order a vehicle and a driverless car would arrive to pick them up at their location.
These examples provide only a sampling of how autonomous vehicles can reduce overall fuel consumption by reducing fuel consumed by vehicles on the road (less traffic) and reducing the overall number of vehicles on the road (ride sharing). 3. 2. 3 Safety “From 2001 to 2009 American roads claimed 369,629 lives. And the culprit was not poorly lighted thoroughfares or faulty gas pedals but us – one landmark study cited ‘human errors’ as the ‘definite or probable causes’ of 93 percent of crashes. ” (Vanderbilt, 2012) Humans 4 Sebastian Thrun keynote speech IEEE International Conference on Intelligent Robots and Systems (Sep 2011) ave physical limitations. While a stereo radar system cannot see around a corner any more than a human can, the 360 degree radar can see cars in a driver’s blind spot or objects and people blocked from a driver’s vision by the car frame. The car’s processing capability provides faster reaction time to unanticipated objects appearing in its path (i. e. a ball bouncing or child running into the street). Taking this thought a step further, these driverless cars can not only perform better than your every day “good” driver, they can eliminate the bad ones. The robot car does not have emotions.
It does not drive while angry or emotionally distraught. It is not distracted by incoming phone calls, email, or text messages. It does not take its eyes off the road when changing radio stations or CDs. And most importantly, the robot car does NOT DRIVE DRUNK OR UNDER THE INFLUENCE! 4. 0 Challenges While the concept of driverless vehicles certainly appears tantalizing among its potential future uses, their reality remains somewhat far off. The autonomous vehicle’s greatest challenge is the dearth of legislation and regulation allowing it to operate legally on the United States road system.
Augmenting the legality perspective, from a social aspect, naysayers among drivers and auto industry leaders generate negative sentiment against this revolutionizing technology. 4. 1 Legislation and Regulation “Questions of legal liability, privacy and insurance regulation have yet to be addressed, and [many suggest] that such challenges might pose far more problems than the technological ones. ” (Markoff, 2012) For instance, when an autonomous vehicle gets into an accident, how should the legal system assign fault?
Does it matter if there are passengers in the vehicle? While the driverless car does bring a vehicular version of the well-known airplane black box that can be programmed to record the duration of a trip or the last 15 minutes prior to an accident, no precedent has been established nor regulation authored to lay ground rules for how the car’s black box can be used in legal proceedings. Hand-in-hand with liability, how and to what level should federal and/or state regulations mandate insurance requirements? O.
Kevin Vincent, chief counsel of the National Highway Traffic Safety Administration (NHTSA) states, “The federal government does not have enough information to determine how to regulate driverless technologies. ” (Markoff, 2012) In fact, of all 50 states, only one has actually legalized driverless vehicles. Although Google has been testing the technology in California, California has not yet passed legislation legalizing them. Working with California’s legal system, Google operates its vehicles based on a finding that robot cars are not expressly prohibited by law from operating.
Actually, in Jun 2011, it was Nevada’s legislators who overwhelmingly approved a bill to pass a new driver’s license bill. The bill puts no limits on where these vehicles can operate and includes city operation in crowded Las Vegas as well as isolated desert highways. The Nevada Sun notifies residents they may see self-driving robot cars throughout the state within the year. (Goldberg, 2011) More recently, Florida and Hawaii have introduced similar bills before their legislators, and rumors exist that California will soon follow suit. (Markoff, 2012) 4. 2 Adoption
Legalities and regulations aside, there are always the stubborn few who resist change. These naysayers cross the spectrum from drivers themselves to government leaders to a minority few in the auto industry to those whose jobs the autonomous vehicle could ultimately replace. Most drivers insist they are good drivers. Why would they turn over driving responsibilities to a computer when they are better than other drivers on the road? How could turning the driving over to a computer be safer than a human in control? Other drivers simply claim to “love driving… too much to ever let a car do the driving! (Wood, 2012) Emphasizing the characterization that a self-driven car could not be as safe as one in control of a human and speaking as a representative of the NHTSA, official O. Kevin Vincent asserts, “We think it’s a scary concept for the public. If you have two tons of steel doing down the highway at 60 miles an hour [ and another] two tons of steel going in the exact opposite direction at 60 miles an hour, the public is fully aware of what happens when those two hunks of metal collide and they’re inside one of those hunks of metal.
They ought to be petrified of that concept. ” (Markoff, 2012) While many auto manufacturers are working to bring autonomous vehicles to the road and are, in fact, working with Google to lobby for the development of appropriate legislation and regulation at both the federal and state level, at least one major U. S. company insists it will not. While GM, Vovlo, Audi, BMW, and others take incremental steps toward automation, Ford CEO Alan Mulally “said quite firmly that Ford would not be developing self-driving cars, or even introducing self-driving mode in vehicles. (Wood, 2012) And finally, what of the people and businesses whose livelihood or at least whose supplemental income is tied directly to driving a car?
Limousine drivers, cabbies, shuttle services, even the diesel bus drivers replaced by Pod Cars at Heathrow? The autonomous vehicle, summoned as an on-call commodity, could even supplant valet parking in a future when cars talk to one another, provide intent, and coordinate to allow a boxed in vehicle to exit. While others may have opinions and others may miss business pportunities, people who rely on driving to pay for room and board certainly have motivation to slow the mainstreaming of the autonomous vehicle. 5. 0 Conclusion and Personal Thoughts This paper summarizes some key points in the autonomous vehicle’s recent history and discusses, in general, the use of artificial intelligence and decision support systems as core elements that make the cars viable. In addition to providing examples of current applications of driverless vehicles, the paper postulates on a smattering of potential future applications.
It is simply not difficult to consider the myriad of other uses that can and will likely develop in the future. While regulations and legalities remain an unsettled ground, safety, fuel economy, and other physical efficiencies demonstrate clear benefits that have already begun to tip the scale in favor of mainstream adoption. In the course of doing my research on this topic, I found myself excited for the potential future offered by autonomous cars. Two of my friends have newborns.
I said to them, “How nice would it be to be able to sit in the back seat and comfort your child with your presence when she is crying up a storm rather than leaving her back there alone, because you must sit in the front seat and drive the car to get you home after the baby just had his/her vaccinations? ” Another friend with four children complains of not being able to juggle all her children’s pickups and deliveries without a second driver. And I thought how a ride-share vehicle that was programmed to pick-up a team of children up for soccer practice could add convenience.
While my parents have not quite reached the age where they cannot drive themselves to one place or another, they are certainly close. The benefits of an autonomous vehicle which gets them from one place to another safely would certainly take a load off my mind. After all, how often do we pass the little old man or little old lady hunched over the wheel of a car weaving in the road or driving slower than the speed limit and wonder how he or she still has a drivers’ license?
Even the disabled could gain a previously not feasible degree of independent living. These examples highlight benefits individuals can derive from self-driving cars. Though I have not yet considered or explored in any depth the business opportunities and public transportation advantages associated with mainstreamed autonomous vehicles, they certainly exist. I must admit, I am in CNET Molly Wood’s camp: Self-driving cars – Yes, please! Now, please!
(2003). McGraw-Hill Dictionary of Scientific & Technical Terms, 6E . The McGraw-Hill Companies, Inc. Breimer, E. (2006). DSS: Decision Support Systems and AI: Artificial Intelligence. Retrieved Mar 26, 2012, from Siena College CSIS-114: Management Information Systems: http://www.cs.siena.edu/~ebreimer/courses/csis-114-s06/lectures/ Brown, J. (2010, Dec 27). AI Autos: Leave the Driving to Us. Retrieved Mar 25, 2012, from Wired.com: http://www.wired.com/magazine/2010/12/ff_ai_drivebywire/all/1 Cunningham, W. (2011, Aug 5). Google driver crashes autonomous car. Retrieved Mar 25, 2012, from CNET: http://reviews.cnet.com/8301-13746_7-20088732-48/google-driver-crashes-autonomous-car DARPA Grand Challenge. (2008). Retrieved Mar 25, 2012, from Wikipedia: http://en.wikipedia.org/wiki/DARPA_Grand_Challenge Goldberg, D. (2011, Jun 26). Self-driving robot cars about to hit Nevada highways. Retrieved Mar 25, 2012, from LasVegasSun.com: http://www.lasvegassun.com/news/2011/jun/26/self-driving-robot-carsabout-hit-nevada-highways/
Guizzo, E. (2011, Oct 18). How Google’s Self-Driving Car Works. Retrieved Mar 25, 2012, from IEEE Spectrum: http://spectrum.ieee.org/automaton/robotics/artificial-intelligence/how-google-self-drivingcar-works Markoff, J. (2012, Jan 23). Collision in the Making Between Self-Driving Cars and How the World Works. Retrieved Mar 25, 2012, from NYTimes.com: http://www.nytimes.com/2012/01/24/technology/googles-autonomous-vehicles-draw-skepticism-atlegal-symposium.html PCMag.com Encyclopedia. (2011). Retrieved Mar 25, 2012, from PCMag.com: http://www.pcmag.com/encyclopedia_term/0,2542,t=autonomous+vehicle;i=57132,00.asp Vanderbilt, T. (2012, Feb). Let the Robot Drive. Wired . Witkin, J. (2011, Aug 5). Pod Cars, Moving Silently at Heathrow’s Terminal 5. Retrieved Mar 25, 2012, from NYTimes.com: