Automated driving is coming! Long live manual driving!
Automated driving is quickly gaining ground (or coming back around, depending on how good your memory is) as the “hot topic” within the transportation community. For researchers in our field, automation prompts a dialogue about the future of transportation engineering, operations, research, and policies. During mid-January, many VTTI researchers attended the Transportation Research Board annual meeting, where the term was seemingly everywhere, from poster sessions to discussions at booth exhibits to conversations in hallways. For the general public, however, it’s a term that causes some confusion and misunderstanding.
Automation itself, at least as it relates to driving, has been evolving for decades, from the availability of automatic transmission 70 years ago to electronic stability control during the last decade to automated braking and steering in current vehicles. Even cruise control is considered a factor of automation.
Automation, therefore, has been a tangible aspect of transportation longer than most realize. And while the goals of automation remain relatively constant (e.g., fewer crashes, precise vehicle control, increased mobility, reduced vehicle emissions), the technologies developed to meet these goals evolve and adapt. Given the multitude of automated systems currently on the market or nearing market penetration, it should come as little surprise that the term automation has no clear-cut definition. Instead, industry leaders and researchers discuss levels of automation, though they continue to debate the ideal way to define these levels. For instance, VTTI researchers are working with industry leaders such as GM and Google to answer fundamental questions about how drivers transition into and out of automated vehicle features classified by the National Highway Traffic Safety Administration (NHTSA) as “Levels 2 and 3.” For the purposes of this particular project, a Level 2 automated car is considered to have several combined functions of automation, while Level 3 is considered to have limited self-driving capabilities.
The taxonomy of automation could take up an entire blog entry, but the takeaway here is that the terms “automated” or “automation” do not necessarily mean a fully self-driving, or autonomous, car. While VTTI has been studying various levels of automation for many years, particularly as they relate to safety, we recognize that a transportation system that solely comprises fully autonomous vehicles will not occur in the near future. I have heard policymakers assert that issues such as distraction, impairment, fatigue, seat belts, and connected-vehicle technology will be past because full automation is just a few years away. However, we must all realize that manual driving will continue to exist for a very long time.
But why? This question spawns several others for debate, several of which I will address below:
1) The technology for automation is essentially proven, so it will happen soon, right?
Actually, fleet turnover for new technologies usually averages 25 years. It simply takes a long time for new vehicles to proliferate the market. The technologies you see in current vehicle models have undergone years, and sometimes even decades, of prior testing of the systems and driver performance, behavior, and acceptance of these next-generation technologies. Quite a few safety systems that now come standard on newer vehicle models were tested at VTTI years before deployment. Additionally, there appears to be no real opportunity for an aftermarket within the field of automation, at least in the near term. Retrofitting legacy vehicles—or older vehicles—with newer technologies that perform to the same standards as new automated vehicles is not likely.
2) Of course, automated driving will be safer, right?
An argument that you hear from many in our field is that drivers cause 90 percent of crashes, thus getting the driver out of the loop will be a huge safety benefit. While true, building an affordable automated system that performs considerably better than a non-distracted driver will be difficult to achieve, at least in the near term. For example, during 25 years of driving, an average driver will execute three million successful braking maneuvers, with one failure. It’s hard to imagine that an automated system would perform substantially better than that in the myriad anomalous situations and conditions that occur within the larger context of driving (e.g., day, night, glare, weather, work zones, urban environments, random road hazards, etc.).
3) Everyone will want to give up manual driving, right?
During 2012, the percentage of manual transmissions sold in the U.S. actually increased (up to 7 percent, which was nearly double the manual transmissions sold during the preceding five years*); this is 70 years after the introduction of the automatic transmission. While there is some truth to the statement that younger generations may prefer an automated vehicle, this is too general a statement to make. An increase in manual transmission sales at least implies there is a segment of the population that prefers to manually engage with the vehicle, mostly because they like to drive.
There is also a marked difference between communities. In California, which has been among the first states to enact legislation designed to regulate automated and autonomous vehicles, commutes regularly exceed one hour. However, the average travel time to work in the U.S. is 25.7 minutes, with approximately 26 percent of these commutes occurring under 15 minutes.** While increased levels of automation may make sense and be popular in some areas such as California, these technologies may not necessarily gain traction within driving populations that experience few travel-time and congestion issues.
There is also the concern that some people will fear automation because of the lack of control. For example, though air travel is common, several studies have found that close to 40 percent of the population experiences some form of anxiety about flying, due in no small part to the lack of personal control. Hence, acceptance of relatively newer technologies that facilitate automation will take time and education. Even then, the acceptance rate of automation will take many decades before it approaches 100 percent.
4) Automated technologies will be reasonably low cost, so everyone can afford them, right?
Last year, Americans bought nearly three times as many used cars as they did new cars. Because the price point is lower, there are still more cars equipped with older technologies on the road than those equipped with newer automated technologies. And while newer luxury vehicles include the height of next-generation technologies, more performance options equal more money. According to Cars.com, 2013 sales of these luxury vehicles remained about average compared to the same period during the preceding eight years. Although none of these cars are “autonomous,” this trend may show that the public is not clamoring for technological advances, which may delay the market penetration of higher levels of automation. My dissertation in 1985 was an evaluation of the benefits and costs of a moving map navigation system (think Garmin or TomTom). While the benefits were clear and the costs were pretty low, it took many years before even these relatively simple devices became commonplace, which may perhaps be a lesson for the future.
All of this is not to say that automated vehicles and technologies are not viable options. There is no doubt in my mind that automated and even autonomous vehicles are coming. However, I also believe that vehicle automation will be an evolution (as it has been for the past several decades) and not a revolution, as some suggest. To that end, VTTI is continuing, and even increasing, our research into the future of automation because we understand that next-generation vehicles and technologies have the potential to increase safety and mobility. However, we must acknowledge that a world comprising only autonomous, self-driving cars is not a reality we will experience any time soon, if at all. Drivers continue to have different preferences about vehicle performance. There are new car buyers who still want just the basics, while there are some who would be more than thrilled with the realities of a fully autonomous car. It’s imperative that, as researchers, designers, and policymakers, we keep moving towards zero deaths and sustainable transportation in a heterogeneous vehicle world. Research and policies that address manual driving (e.g., alcohol and other drugs, crashworthiness, distraction and attention, fatigue, driver error) will continue to be very important. How newer (automated) and legacy (manual) vehicles interact as part of a larger system to increase safety, improve mobility, and decrease environmental impact is where research efforts need to continue to be focused.
Automation is a fascinating topic with multiple implications and divergent opinions, which is why we’ll be back with more installments about this issue. Stay tuned…
**2012 American Community Survey, U.S. Census Bureau