Infrastructure 2.0 is Smart Everything
Every aspect of roadways needs updated and improved.
Roadways as a public concern didn’t exist in any significant way prior to the automobile. Back then, roadways were mostly compacted earth that was formed naturally in response to horses and wagons traversing the same area repeatedly, like a footpath forms through a lawn. In cities, high traffic roads would be improved on occasion with slate tiles, cobblestones, brick, or logs. Sometimes in larger, more organized cities this work would be performed by a public agency, but just as often it would be done by a group of interested citizens or by the property owner that fronted the road or owned the land the road was on.
Cities and states planning and funding roadways is a relatively new practice in the scheme of human civilization. We forget that developed, dense cities are still very new, for example Manhattan doesn’t have alleys whereas Chicago does, because Manhattan was laid out on a grid before city planners realized how important alleys are to city management. Boston — like most European cities — doesn’t even have a grid because it grew up before city planning was a thing. It wasn’t until the automobile started to become commonplace that the public was expected to absorb the burden of constructing and managing infrastructure as a standard practice.
Most of the rules about how to plan, fund, design, and build roads were established the 1920s and 1930s. By the time of the Eisenhower Interstate System after WWII, the “way we do things” was set in stone by City Councils and Departments of Transportation. While these rules and regulations have been refined since then they haven’t been seriously updated. Even as we moved from paper to digital, our public agencies have barely budged. That has excluded incredible advancements in the state-of-the-art of road building that have been discovered over the last 100 years.
We now have the ability to do things that weren’t even dreamed of when the auto was invented. If we’re going to be investing trillions of dollars in public infrastructure, the first thing we need to do is update our expectations to the modern era — we can’t build roads for the 2050s based on requirements written in the early 1900s by people who were still reeling from the shift away from horse-drawn wagons.
Smarter Materials
There are lots of materials that we could use for road building that improve significantly on asphalt and concrete. These include CO2 absorbing concrete, tailpipe pollution absorbing concrete, pervious concrete, extremely durable and long-lived geopolymers, non-metallic reinforcement products, coatings, and admixes — improvements that can have dramatic effects on the pollution inherent in producing pavements, the ecological impacts of tailpipe emissions, and hard impermeable surfaces covering much of the water table. For the most part none of these materials are used because they’re more expensive, and I’ll talk more about how procurement issues limit us in a future article.
But while better materials are important and absolutely need to be adopted, improving materials is like Infrastructure 1.1. Materials alone can’t get us to Infrastructure 2.0 any more than better pens and paper could get us to computers — yes, it’s better than what we had, but it’s not the dramatic leap forward that we need for the next generation of infrastructure.
Smarter Building Methods
Over the last few decades construction has tended towards prefabrication of materials — producing the building blocks off-site in a factory, trucking them in, and assembling them onsite. Prefabricated skyscrapers have set records for construction speed and reduced cost by 30% versus cast-in-place concrete. But you have to get economy of scale to get those savings, and the way we build roads has slowed widespread adoption.
Particularly in very expensive and relatively unsafe operations like large bridge, overpass, and fly-over elements, precast concrete has been able to compete. And in some high-traffic states where road closures are very expensive, precast concrete pavement has been used for almost 20 years in limited doses, with California and New York being some of the leaders in these methods.
We also need modularity so that as many roadway sections as possible are designed to be identical and can be mass produced like Lego blocks. These modular roadway sections need to be removable so that they are easily replaced, and so that underlying utilities can be accessed. Most city streets are criss-crossed by power, water, sewer, gas, steam, and data lines laying a few feet under the surface. We can’t justify investing trillions in rebuilding America’s roads if that means we can’t get to the utilities underneath that also need replacement, repair, and maintenance. And if those utilities don’t need servicing now (protip: most of them do in fact need servicing now), they will need service at some point over the next 35–50 years, so the roadway needs to be removable and utility-accessible.
Smart Systems Built into the Roadway
Until now we’ve been talking about incremental improvements like changing the wax or the wick of the candle, but integrating smarts into the roadway is the light bulb moment.
Data Collection
Roads need to collect data about traffic — not personal information like “John Smith drove from location x to location y at z time at n speed.” The data roads need to collect is generic metadata like “389 cars turned left at this intersection between 7:30am and 8:30am on April 4th, 2020.”
Imagine if a website only had web traffic data for a few days of traffic, collected every few years. The website owner’s ability to understand user behavior would be miserable, and they would be seriously limited in their ability to observe and respond to trends, changing needs, and users interests. Right now, traffic data is collected once every 2 to 5 years through highly manual methods like rubber hoses attached to counters that collect a few days worth of data.
There are more advanced methods, like connected signal systems or cameras, but they are uncommon, relatively expensive, produce poor quality data, and are awkward to use. Even when used, these systems don’t feed information to a database that is easily accessible to anyone other than the public agency. And even when that happens, there’s not a way to aggregate data between and across agencies.
Because of these limitations, most commercial users have turned to GPS signals from smartphones, but these GPS-based systems aren’t that great either. As shown recently from the children’s wagon prank, systems relying on GPS can be spoofed by loading a ton of cheap used phones into a wagon and placing it at a major intersection. Then there’s the “urban canyon” effect on GPS among tall buildings (like NY and SF where congestion is the worst), the location-error problem (What lane is this car in? What direction is it going? Is that a slow moving car or a pedestrian? Is the pedestrian in the street, on the sidewalk, or in a building?). GPS to get traffic data is like a cordless phone — an improvement over the wall phone in the kitchen, but also not the revolution of a smartphone in your pocket.
A significant improvement is to build the traffic data collection systems directly into the roadway itself, so that from the time it’s installed, the road is constantly collecting (anonymous, non-user-specific) traffic data. This data describes vehicle locations, counts, speeds, weights, directions, lane usage, and other rich, detailed analytics, then automatically feeds it into a cloud database that aggregates it with data-rich traffic data from a multitude of other locations — proximate and distant — and provides a simple to use analytics and informatics dashboard.
Communication Systems
When I get into a Lyft, I often ask the driver “What’s something that your riders want that you don’t have for them?” The most common answer is “Wi-Fi”. Many newer model vehicles have Wi-Fi hotspots built in, but these all feed back to limited and relatively expensive 4G connections. Connected vehicles need some kind of backbone network access to work, and so far most of the concepts around CV operation is to use other CVs in the area to bounce messages, which is a broken model in countless ways. And when we consider the demands of autonomous vehicles, they need ubiquitous high-speed connectivity for a multitude of reasons.
The vast majority of wireless service users are within a few hundred feet of a road, and emerging networks like 5G are reliant on communication systems directly in the public right-of-way. High-speed 5G frequencies work more like Wi-Fi than 4G, in that Wi-Fi has a relatively small service area while 4G provides coverage over a several mile radius from a tower. Millimeter wave 5G will only work for about 300 feet / 100 meters (ish), so you have to have antennae all over the place, whether in cities or along highways.
Deploying on buildings is a headache because you’d have to negotiate leases with thousands of individual property owners to get coverage throughout a city. But using public assets isn’t a solution because municipal-owned poles aren’t ubiquitous, aren’t positioned in the right locations, and are heavily regulated. Cities don’t like having 5G antenna systems all over the place, and have charged outrageous permitting fees for relatively small 5G installations because of the aesthetics. AT&T went to the FCC to demand that the FCC order cities to give cost-free access to carriers for 5G, and the FCC basically told AT&T that it wasn’t the FCC’s place to demand that of cities.
Even if you can get the antenna placements figured out, you need cabinets all over the place to stash the control systems, and you need backbone fiber.
That’s where Infrastructure 2.0 comes in. By integrating the deployment of fiber backbone, front-end antennas, utility cabinets, and power systems into the deployment of the roadway project, and hiding the antennas directly into the roadway, you can slash the cost of installing 5G. The outcome of this kind of communication system integration is significantly lower costs for deployment, improved internet backhaul, improved wireless fronthaul, invisible antennas, and brand new roads. It’s a perfect match to what cities need right now, and supports the emerging needs for connected and autonomous vehicles in the near future.
In-Motion EV Charging
Tesla has been on a tear the last 5 years and EV adoption is greater than it’s ever been. The Supercharger network has been a great way to address “range anxiety” for EV adoption. However if we look at last Thanksgiving, there were hours-long lines at Supercharger locations. And Tesla has only sold 1m vehicles so far, how are we going to handle all the charging demand when most vehicles are EVs?
And what are commercial fleets going to do? A big-rig can’t stop every 200 miles to spend 30 minutes charging. We can’t really make the trucks significantly more expensive (and heavier) to get more range. Widespread EV adoption needs a different model than “plug and wait”.
The solution is in-motion EV charging, by integrating wireless charging into the roadway so that EVs can charge as they’re driving. This reduces the size of the battery in the EV, reducing its weight and cost while extending its range, and eliminates range anxiety by making the vehicle’s effective range the range it can travel when it’s not on an EV-charging road.
In-motion EV charging solves range anxiety for commercial fleets and private drivers in a way that doesn’t result in long waits at a charge point. Infrastructure 2.0 has to include in-motion EV charging in order to drive more EV adoption and produce the positive climate impacts we need.
Navigation for Autonomous Vehicles
The current mindset for autonomous vehicles is to make the vehicle responsible for mapping the environment, figuring out where it is, identifying all the other vehicles, and navigating its path. This is an extraordinarily difficult task, as we’ve seen from Tesla’s continued failed promises to deliver “full self-driving”. It’s getting better at a glacial pace and is way behind the promises we’ve heard time and time again.
Waymo is behind schedule too, providing extremely limited services in Arizona to a tiny set of pre-selected people at very slow speeds. Cruise was supposed to debut full autonomy in 2019, but finally admitted that its vehicles can’t make a left-hand turn reliably. Uber’s effort got thrown into a legal grinder that Anthony Levandowski is still trying to get out of.
As it turns out, putting the full responsibility of mapping, positioning, obstacle identification, and navigation onto the vehicle is an extraordinarily hard task. (Who’d ever have guessed!?) But what if we didn’t have to? What if we had something built into the road that could identify the position of all the vehicles in the area (see Data Collection), overlay it onto a map of the area, then communicate that information to all the vehicles that can use it (see Communication Systems), so that the only thing that the vehicle had to do was identify pedestrians and navigate? Wouldn’t it be far, far easier if we minimized the stuff that’s happening inside the car and shifted everything else out into the network?
You know… like how literally every other network that we’ve built works! The cell network doesn’t depend on cell phones, the internet doesn’t depend on personal computers, so why in the world did anyone assume that autonomous cars would depend entirely on the car itself to work, instead of the car being a “thin client” device attached to a powerful network?
With Infrastructure 2.0, we have the ability to shift the really difficult tasks for autonomy into a network that is far better equipped to accomplish them, and reduce the workload for the car to the things that the car is best suited to doing. And that brings us to our next topic, edge computing.
Edge Hosting
Intersection signals are controlled by a control box, typically green, that sits near the intersection. When we’re adding all these sophisticated new sensors, antennas, chargers, and other capabilities to the roadway, we also need to add edge computing assets to manage all the services.
Your cell provider already uses some edge computing and storage to manage, for example, content distribution for Netflix. If you watched “Tiger King” on your phone, it’s probably not coming from a datacenter in Virginia, but from a copy that’s stored at a server at the cell site that your phone is connected to. This means that the file only has to be transmitted over the wireless link to your phone, instead of traversing the entire network. Edge hosting makes the network faster, more reliable, and significantly more efficient.
Smart Infrastructure needs edge hosting for data collection, operating the distributed antenna systems, providing connectivity services for connected vehicles, managing charge systems for EVs, mapping services for autonomy, and for managing content distribution of all types across the network.
Technology Isn’t the Only Improvement We Need
So far I’ve been describing how new technologies can help us with Infrastructure 2.0, from materials to design to implementation, and all the new features and services we can get from including “smarts” in our infrastructure.
But technology by itself won’t do the job. We can’t keep using the same outdated, insufficient systems we use to pay for and manage roadways to implement all these new technologies, because the ways we pay for and manage roadways today actively (though unintentionally) prevent improvements from being adopted. If we’re going to make it to Infrastructure 2.0, we also need to reconsider the way that we procure roadways, how they’re managed, and how they’re funded.
In the next article in this series, I’ll talk about how we can update our procurement, management, and funding practices to accommodate and incentivize the adoption of Infrastructure 2.0.
