Anthonares

Each year the nations highways grow more and more crowded, especially in major cities. And with those crowds come accidents that wreak tremendous economic and social damage on our nation. Despite tremendous efforts by governmental and private sector agencies, ridesharing/carpooling programs have been largely unsuccessful, and public transit ridership has not increased in over 30 decades. More people are driving than ever, and like it or not, Americans will not change these habits any time soon. The environmental effects of all of this driving can be largely mitigated today (with low emissions vehicles, coupled with the purchase of offsetting carbon credits from Terrapass), but to decrease traffic jams and accidents, we need automated highways and self-driving smart vehicles.

People always look at me a little skeptically when I claim that we need automated highways (I’ll explain exactly what I mean below), so first let me justify exactly why people are totally unfit to drive for themselves.

Crowded Roads, Unsafe Drivers
Even though enormous arterial highways with 6 lanes in each direction are not at all uncommon, Americans spend 62 hours stuck in traffic, on average, each year. Driving has become a less convenient means of transport over the last few decades for a number of reasons. Rising home prices and changing priorities have meant that more and more of us live in the suburbs and exurbs than ever before (in fact, the word exurbia was invented because of the longer commutes). Since 1983, the average commute length has increased by 88%, from 8.5 miles to 16 miles in 2005. And, because of the enormous costs of building new mass transit (light rail, subway) lines, these systems have not been able to keep up. The average annual number of boardings on a mass trasnit system per capita decreased during the approximately period (1980-2002), from 37.4 to 30.6.

As I discussed in my last entry on the Physics of Traffic Safety, the per capita rate of motor vehicle deaths and total deaths has been on the rise in this country for the last decade while other industrialized nations have continued to see decreases. In 2004, 42,636 people lost their lives in traffic accidents. National Safety Council estimates lead to a total cost to the US economy of $222 billion, just for those accidents that occurred in 2004. As discussed in the previous entry, a big part of saving lives in accidents will come from improving seat belt and airbag design along with making light trucks and SUVs share the road more safely with other passenger vehicles.

While traffic jams are largely an urban phenomenon, traffic deaths are not. Rural areas account for only 40 percent of vehicle miles but nearly two thirds of all traffic deaths. The reasons why include the relative isolation of such accidents, the high speeds at which they occur, and lower safety belt use. Also, the National Highway Transportation Safety Administration reports that drowsiness is suspected in 29,000 fatal crashes per year, which is anecdotally suspected of being the leading cause of vehicle deaths. Beneath that, we have alcohol consumption which accounted for 16,694 deaths in 2004. In fact, driver error is suspected in 95% of all crashes, so human error killed approximately 40,000 in 2005 alone.

Smart Cars and Automated Highways: A Technological Solution
Automated highways, combined with smart vehicles, and improved safety systems could save almost every one of these lives. No transportation system even approaches vehicles in terms of customizability and mobility, and Americans have thus far shown no inclination to change their behaviors around mass transit schedules (or walking/biking to work, but that’s another topic altogether). If we insist on driving, and in so doing threatening the lives of others and ourselves, it should at least be done safely. And to do that, we need to get humans out of the process as much as possible.

Automated highways are years away, even though the component technologies seem mostly in place, but surprisingly the smart vehicles are nearly already here. Honda announced this week that it is marketing the Automated Driver Assist System (ADAS) in its Honda Accord. The ADAS combines the standard cruise control speed-adjusting radar system with a lane-sensing camera on the back end. The system operates only on freeways and four lane roads, but it allows the driver to turn on the system and have the car do the steering, even at highway speeds. Every 10 seconds it beeps at the driver if they are not holding the steering wheel. GM is planning on releasing a similar, but more advanced, system on its Opel Vectra in 2008. Eventually, I envision smart vehicles as being capable of sensing the complete environment around them, as well as mapping the road ahead to detect any dangers. These vehicles will also have systems to alert the driver when attention is needed, and to monitor the driver for drowsiness.

Smart vehicles will save lives, but automated highways will save time and increase capacity to handle future demand. What I mean by automated highways is a system that consists of three things: 1) an anonymous system for GPS coordinate reporting that tracks a single-instance vehicle-ID, 2) a network of interconnected road monitoring computer systems that provide instructions to the smart vehicles on what turns to take, and when to change lanes, 3) a mechanical failure/accident detection and reporting system to drastically reduce traffic jams and emergency response times. We could implement the first component today. GPS is accurate enough when coupled with smart vehicles for the purposes of this system. Privacy can be maintained through the use of transparent open source code and regular independent auditing. In no case will a driver’s identity be attached to a vehicle GPS-ID.

The second and third components will take a little bit more time. I was at the University of Illinois in Urbana-Champaign a few years back, and I watched a demonstration of a system that routed four radio-controlled vehicles on a 20ft.x20ft. board with painted roads and parking spaces. A network of computers controlled the vehicles, and enabled them to avoid collisions and reach their destinations. But with alarming frequency, the vehicles left the road, and small perturbations of the system led to large deviations in behavior. This type of phenomenon is known as chaos. Traffic systems, with millions of independently moving (and interacting) vehicles is fundamentally chaotic. No computer system devised could stably route the entirety of US vehicle traffic at reasonable speeds. However, over smaller areas, and in good conditions, the system is locally stable, if still highly nonlinear. Therefore, a network of monitoring systems each controlling a small patch of an automated highway would be capable of managing the routing of the vehicles.

But, most of the actual driving needs to be done “on the ground” using a suite of sensors that are independent of the automated grid. The nice thing about this type of dual-layer system is that were the automated grid to fail, the vehicles could still drive safely, and were the vehicle’s systems to fail, the passenger could take over.

Rising population will make automated highways a reality simply because humans are completely incapable of safely driving in such crowded conditions. Increasing in-cabin distractions and longer commutes will make both drowsiness and distractions a larger cause of vehicle deaths as well. Smart cars and automated highways will largely remove us from the loop, and in the process convey us safely and more efficiently to our destinations. And surprisingly, the smart cars are only a few years away.