Singularity Watch
HomeNewsletterReading GroupsConferencesPublications"Singularity Studies" LinksDegree ProgramsCritiques


Underground Automated Highways (UAH)
for High-Density Cities:
A 2030-2060 Forecast

Three Technologies

Convergence Vision

Urban Greenbelts





NOTE: The ten-page professional version of this forecast is available here:
Underground Automated Highways (UAH) for High-Density Cities (PDF), John Smart, 2005.


The Coming Convergence of Three Technologies: A Post-2030 Developmental Attractor

tunnel boring systems

automated highway systems

zero emission gas-electric vehicles

What will urban transportation look like after 2030? Which major new features of our future transportation infrastructure are highly likely to emerge in coming decades? Developmental futurists, including several of us at the Acceleration Studies Foundation enjoy confronting such important and multifaceted questions.

Will we primarily see, as technology scholar James Burke calls it, more "Business As Usual," e.g., simply more of what exists today? Will there be a profusion of double- and triple-decker freeways? Will we have more stylish and efficient mass transit? What will be the disruptive new advances? Which particular emergences point the way to our next great metropolitan transportation paradigm, and why?

It doesn't look like it will be Personal Rapid Transit (PRT) that beautiful but utopian top-down designed system of fixed-track "taxi cars" that has failed everywhere it has been attempted. PRT involves too many compromises of movement freedom by comparison to our present bottom-up evolved, out of control amorphous network of cars and roadways that that make it easy for us to drive exactly anywhere we want in physical space. Furthermore, our Moore's Law-aided cars of tomorrow will be doubling their navigational intelligence every year forward over our existing road infrastructure, making today's roads incrementally safer every year forward. Thus Moore's law applied to navigation will make it exponentially easier for tomorrow's smart cars to compete against any fixed-track alternative proposal. At the same time, new tracks themselves have become too expensive to construct over the roadways of our most desirable cities, even if we could ignore their visual blight, which increasingly we cannot.

So what will be the shape of tomorrow's dominant urban transportation system? Post-2030 it seems likely we'll see the beginnings of an underground automated highway (UAH) network emerge, one that may extensively undergrid many of our major cities by the late 21st century. The degree of emergence of this new system may depend on the way a number of important local contingencies play out, including our own insight (early or late) into the unique advantages of this system, and the strength of our political resolve to set up test systems (again, early or late, the choice is ours).

To make our case, let's look briefly at recent and coming developments in three important technologies: tunnel boring systems, automated highway systems, and zero emission fossil fuel and fuel cell vehicles. The convergence of capacities in these three technologies may create a transportation option that looks much more resource and capacity optimized than any other 21st century solution that has been proposed.

1. Tunnel Boring Systems. Digging connector tubes between dense urban cores underneath cities, if we can complete several such pilot projects some time after 2020 (earlier would not be practical, for reasons explained below) would show how easy it is to create major improvements in our space, time, energy, and matter efficiencies (STEM efficiency) of social interaction in our largest, most population dense cities. Compressing social interaction time is what cities do best, and it why they will continue to outcompete rural areas in living choice for most first world citizens, even as rural homes become ever more affordable as getaway spots for city dwellers.

As Fred Hapgood has noted ("Sub-Urban Renewal," Wired, April 2003), our tunnel boring technologies have dramatically improved since 1994's completion of the Chunnel, the 31-mile rail tube that connects France and Britain. These technologies were impressive then, and they've become increasingly intelligent since. If you recall, laser-guided borers (reflected back to the surface with mirrors) kept the Chunnel's digging teams to within an error margin of only a few feet when they finally made their connection under the ocean floor. He notes that in only nine years since, several components of the digging process have become highly or fully automated, with teleoperated machines laying reinforcements, and with many of the human operators remaining above ground, away from the dangerous rock face. Costs for these digs are dropping dramatically as well.

Mike Smith at, the leading portal for the tunnelling industry, notes that the recently completed seven-mile Flam-Gudvangen tunnel in Norway, working in ideal geology, cost only $1.50 per cubic foot, for both excavation and construction. If this was like similar projects, the majority of construction cost came in excavating the tailings. But once we have one surface-level lane available to automated trucking systems (see below), that cost will fall to a fraction of its present value. As our technologies and automation continue to improve seems likely therefore that prices will eventually fall below the current cost of many of our present above-the-ground highway projects, particularly once we factor in the high social cost of losing highway access during the extensive and the permanently ongoing road construction that occurs in any growing city.

Certainly we've had this vision for a long time. Consider this Newsday article from 1999, "Tunnel Vision Looking Dim," on the failed dreams of suburban tunnels in the 20th century. While all of this is true, it doesn't deny that the dreamers are always seeking ways to create compelling realities, as with Boston's "Big Dig." Albert Appleton's quote seems particularly prescient: "There's going to be a real revival of urban tunneling. [On one hand,] we need roads and we need subways, and on the other hand, neighborhoods are not willing to be torn up the way they were in the '50s and '60s."

Where will we put all the dirt (tailings) from these excavations? That is an excellent question, and it reveals one of many hidden constraints on this highly optimistic vision. Large scale underground tunnels would only be economically feasible to construct once we have at least one automated highway (AH) lane running in a network on our surface highways, giving us the ability to automatically truck the tonnage out to distant landfills. This appears to fix the date for the first such system sometime after 2020, which is the earliest that realistic observers expect to see any deployed and working above-ground AH systems.

In the Wired article, Hapgood suggests that we'll see quite a lot of our present construction dive underground: roads, parking lots, downtown hotels, warehouses, even residential artists' lofts. But before any of that could happen we need to realize that transportation would have to be built first in the above set of possibilities. Otherwise, zoning laws will very effectively shut down new construction in any instance where it might further load our already stressed transportation infrastructure in major urban cities like Los Angeles, a living space that stays self-organized at the edge of chaos.

In other words, transportation is the bottleneck to building in the third dimension, and reaching the kinds of STEM efficiencies Hapgood envisions will happen or not based on our ability to build the network described here.

Let's see what we are up against.


2. Automated Highway (AH) Systems. Some time post 2020, possibly as late as 2040-2050 in a more pessimistic analysis, we shall begin to see special, above-ground lanes with markings that allow our increasingly intelligent cars to self-pilot over long distances. The first automated highway networks will be built when our car's onboard computer, sensor, and actuator systems can solve the "2D" problem of autorouting moving vehicles within special pre-mapped (and realtime updated) virtual spaces. It is reasonable to expect that this problem will be solved many years, perhaps even decades before we have the ability to create autonomous, agent-based air traffic control systems in 3D space for tomorrows airplanes, as this latter is a significantly more difficult computational problem. Furthermore, the vastly increased numbers of humans traveling on (and dying on) today's roads vs. in today's skies makes the former a much higher social engineering priority.

At first, there will only be a few of these special AH lanes, probably taking over the inner HOV lanes on our freeways (first for intercontinental trucking) and later, larger boulevards. For at least the first decade or two, we should also expect that humans will be required to remain behind the wheel, at least theoretically being able to take over this "cruise control" at a moment's notice.

These systems will certainly be built out slowly due to very high initial coast, beginning some time after 2020, between and within our more sprawling and high-density metropoli, like Los Angeles and New York, and at significant road real estate expense. These smart small trucks and vehicles will be carefully-watched curiosities for several years, and automatic collision-avoidance systems will be the interim technology, likely appearing in every new car before we see the first full scale automated vehicle (AV) models (Toyota? GM? Xeon? We wait with anticipation...). When hooked into an AV/AH system, 21st century multitaskers can simply do a lot more living than they can when their hands, feet, and eyes are constantly chained to the road. Being allowed to view video (or the interactive internet of 2030) while driving in an AV may be a killer app for productivity and free time, one that by definition causes many of us to purchase such systems.

The future of an idealized transportation network would include fully automated routing to our destination, allowing us great nap, entertainment, or private time in the increasingly comfortable mobile offices and entertainment spaces that we've been busily building toward since the birth of the automobile. Our early luxury furnishings gave rise to station wagons, minivans, and SUVs.

All roads lead us toward the increasingly magical "swiss-army cars" of tomorrow, able to increasingly dynamically reconfigure themselves to fit the changing needs of the occupant.


Chuck Thorpe's

"Smart Road,
Smart Car


3. Fuel Cell, Electric, and Zero Emission Fossil Fuel (ZEFF)Vehicles. Sometime circa 2030, by many accounts, we'll see the first mass-market affordable fuel cell vehicles. GM's Autonomy frame, perhaps the most provocative design to date, is pictured to the right. As today's high school students will tell you, these beauties produce only water in their exhaust and are really cool.

At the same time, we'll see more electrics, more clean burning natural gas vehicles, and the takeover of the gas burning market by hybrids, which are today almost as efficient as idealized hydrogen cars but without all the problems of conversion, as Dominic Cera notes.


But at the same time, it presently seems highly likely that the majority of our vehicles in 2050 will still be burning gasoline.

The more things change, the more some things stay the same.

Despite what some futurists will tell you, gasoline (whether made by oil extraction, oil shale, coal gasification, coal liquefaction, natural gas conversion, or some other source) probably has a long future lifespan. Fuel cells are nice, but there's every reason to expect that as they are improved, they'll simply be coopted for use in second generation gas-electric hybrids, not lead us to a hydrogen economy, as alluring as that sounds.

Expect to see increasing numbers of Zero Emission Fossil Fuel (ZEFF) vehicles on the market in coming decades. Unfortunately CO2 is not part of this equation, but as we move a more efficient fossil fuel (natural gas-driven and coal gas-driven) electric grid, CO2 sequestration at the power plant will make increasing sense in coming decades, delivering the eco-friendly machines we all seek. Big Oil is not going away, but will continue to innovate heavily in coming years.


Ford's ZEV

Self-driving cars for everyone on the planet who wants them, operating on some future evolutionary, not revolutionary generation of today's power systems seems to be in the cards for at least the first half of the 21st century.

The Convergence Vision: Robust Underground Automated Highways (UAH)

Because of the steadily decreasing price/performance ratios of our tunneling technologies, the increasing intelligence and cleanliness of our automobiles, and the ever increasing social value of our spare time, some overcrowded city (Shanghai? Osaka? LA?) will eventually decide to put a nice, long AH express tube underground, creating a fast connection between two important and yet informationally different urban cores, at the same time bypassing the most gridlocked sections of the city. Only certified, ZEFF, electric or fuel cell automated vehicles will be allowed into these underground connector tubes. Want Speed? Get Intelligent and Go Green.

The first tunnels will likely prove so popular with busy consumers that they'll be sure to upgrade to cars that can operate in them, and in the space of a decade following, leading cities will rapidly turn it into a network. Given the time frames discussed above, I'd expect the first AH "test tube" to arrive some time between 2020 (most optimistically) and 2050 (most pessimistically), in some enterprising city around the world. The choice is ours to accelerate or delay, but not to prevent, as far as I can tell. Combining intra-city AH, ZEFF/fuel cell vehicles, and semi-automated tunneling systems seems like an inevitable developmental attractor for convergence of these three revolutionary technologies.

is Faster!

One of the hidden benefits of eventually putting most of our automotive transport underground is that AH "riders" don't care that there's no scenery. They won't looking around at their surroundings, so they don't need to be above ground. As soon as the cruise control is engaged, if legal, they will retreat to the cozy environs of their cars, buses, and mini-trains, getting on with the business of their lives.

Those Cisco routers under their hoods will handle the merges and interchanges, while serving up the latest information feeds to the passenger compartments. Imagine that: data and physical packets, automatically flowing along on one convergent protocol. Tomorrow's urbanites, if the most fully realized AH networks emerge, will be thus able to "teleport" themselves wherever they want throughout the megacity, with little disruption of the flow of their lives in the process.

Typical AH networks will involve some "fan" of exits to handle the outflow volume, and the more popular underground AH becomes, the more underground parking lots will be built. The strongest surface traffic reduction would emerge when all the city's major above-ground parking structures are directly linked to the tunnel network. Drivers will still have to come back to their seats to drive that pesky "last mile" on the surface by hand. If they don't do that, they will be auto-routed to some expensive city-owned underground storage lot, you can be sure.

Only a very robust A.I. could take over an open-course, surface street driving function, and that eventuality, as we will discuss, changes the nature of the game. But within the first AH network, wherever it emerges, 21st century Jetsons will be zipping along in their vibration-stabilized wombs at seventy to ninety miles per hour, nonstop, nearly as straight as the crow flies. That would make the pleasures and possiblities of the entire New York Metropolitan Area as accessible as the corner store, any time of day.

Want to drive the whole way manually instead? There's always the surface option.

There are lots of other details to confront, of course. Earthquake-proofing of the underground network will have to be done in an economical manner. In the same way that modern skyscrapers are often built on a "fluidized bed" of rock that acts like a stabilizing lake in case of underground movement, we'll have to build something equivalent into our AH around the fault zones, and this will increase the cost, perhaps delaying its emergence another few years. But in the Kobe earthquake in Japan, underground buildings were the least damaged, as the structures moved with the land rather than whiplashing. It's all quite doable.

As no petrochemicals are being belched from these vehicles, underground air quality and ventilation challenges will be far easier engineeering issues than in our present tunnels. Such vehicles may also be required to contain their own independent and emergency air systems, fit into very small spaces by clever auto designers. Rather than escape exits at regular intervals, it seems more likely that automated successors to today's tow trucks would become the "immune system" that AH networks would rely on in case of mechanical failures by cars within the tunnels. Again, these are worthy challenges for our future engineers.

Parking Lot Arboretums and Trailway Takebacks: Growth of Urban Greenbelts

One of the more successful features of Boston's Central Artery Project (the "Big Dig"), is the 40 acres of new park emerging in the Charles River Basin. The parkland is perhaps the greatest consolation to Bostonians for its ridiculous cost overruns and shoddy construction (it's contracts, awarded on a cost plus basis, were a corrupt fiasco by comparison to the efficient megatunneling projects we have seen in Europe and Asia during the same time).

Take out a freeway, and add strips of new park land in its place. That's a powerful combination.

Moving many of our present parking lots and even some of our current multi-level parking structures underground would be another very valuable addition to the AH system. If construction of underground parking became as cheap as underground tunneling, cities would jump to build underground lots for AH-equipped cars. Some residents might choose to store their cars permanently underground at their local lot, paying rent, and freeing up their residential garage space for extra living rooms. Some of these individuals would also loan their stored cars to car "co-ops," sharing multiple individuals per car. Of course some descendant of today's leading rental car companies would offer clever new ideas in this regard as well.

Now we dare consider that if our AH networks can become particularly successful prior to the singularity, much of today's above-ground parking infrastructure, and even some small fraction (10%? Perhaps more?) of our denser city's current roads and highways, are very likely to dive underground as well. Of course, if that occurred, many of today's cities would rezone some of those reclaimed areas as greenbelt, bringing beauty back to some of our most blighted urban environments. I'd certainly love to see a few of our 12-level parking lots disappear, with the ground below them transformed into arboretums. If the economics allowed it, that could be an irresistible image to a 21st century urban voter: the city's less strategic, low-traffic pavement giving way to parkland and new running, biking, and riding trails, while our average urban drive-times substantially decrease simultaneously.

Such foresighted "takeback" (a wonderful eco-term that more of us should be aware of) could occur in a significant manner only after we'd first solved our current rush hour transportation gridlock via a cleverly routed, redundant UAH system. We'll have to do this while accomodating a steadily increasing future city density at the same time, of course. While working through thorny labor issues, and fighting all those famous cost overruns and political logrolls. In other words, there's a whole lot of hard work ahead before an effective and affordable first-generation underground AH network will be realized.

Post 2020, 60% of the world's populace is expected to live in the confines of a city. These magnificent environments are the final destination for human beings prior to the singularity. With robust UAH networks, the transportation times within our great cities could be made better than ever, in a way most residents would find appealing.

While we're on the subject of visions, what can we expect to happen with monorails, that elevated skyway image that 1950's futurists were so excited about?

As much as I like Disneyland and Las Vegas, I think monorails are going to be as rare as double- and triple-decker freeways, double-wide highways, and any of the other visual irritations that our conventional wisdom has offered up for the future of urban transportation. Tomorrow's zoning boards and the increasingly better organized foes of urban blight will very effectively keep such projects from becoming further affronts to our already polluted urban visual landscapes.

It's time to clean up our technology, and, as Mark Wieser says, start making it invisible.

This said, I do believe monorails can be beautiful, and will gain increasing acceptance in a growing number of cities as one element in a plurality of transportation options, as they become increasingly quiet, energy efficient, and to some extent at least, aesthetically appealing. Both high speed elevated rail and mag-lev installations are making steady advances, and both now offer peak operating speeds around 250 mph (likely to be the approximate maximum for many years to come, due to high velocity turbulence).

Such systems remain mass rather than personal transit however, removing both travel environment customization and door-to-door convenience. They will likely only be used in a few high density corridors, and unless they are to become a visual blight, are not likely to even double a city’s current traffic capacity, much less improve it by an order of magnitude. Nevertheless, monorails seem particularly attractive as city showpieces, traveling over the most beautiful areas of each city, as extensions for existing mass transit networks, and as adjuncts to high density tourist attractions.

But for the bulk of human urban transport then, instead of monorails, I think we're headed for mole holes, and later, nature trails. If we decide to repurpose some of our surface roads intelligently enough, we may even see some functional trail networks emerge for that increasing fraction of urbanites who would like the option to use human-powered bicycles, electric bikes, and other low-speed electrics, such as Segways, for some of their local transportation.

Cruising through the shrubberies at a computer governed 15-30 mph, getting a workout if one wants, and enjoying the view all the way to our destination. Actively toiling toward a coming era of human-surpassing machine intelligence, but in the meantime making the best, most humanizing use of the present technology and ingenuity at our disposal. Sounds like a sufficiently green future to me.

In Conclusion

The first links of a coming underground automated highway system, likely appearing circa 2030 in our largest cities, appear to me to be statistically highly probable event, even given the rapid growth of digitization, e-commerce, and virtual reality technologies as increasingly effective surrogates for physical transportation in coming years.

Transportation planners, I'd start asking how we might accelerate the arrival of such productivity and quality of life enhancing networks in our largest cities.

And once we have effective UAH in any city, perhaps the greatest dividend, after the increased social computation and convenience, would be a functional new greenbelt built on the skeleton of some of the surface roads we could now afford to repurpose.

That latter emergence would require particular foresight on the part of the communities instituting these networks.


We at the Acceleration Studies Foundation hope that we will realize such visions in many of our biggest cities, in a balanced and cost-effective manner, sooner rather than later. Will we see the very first link of a successful underground automated highway project completed in 2030, in 2040, or in 2060? Just as importantly, which country will lead the way? The shape of this future is ours to choose. Safe travels to you!

Feedback? Send us an email at mail(at)