Autonomous Vehicles and the Future of Transportation

This semester I enrolled in Autonomous Vehicles and Local Government. Our class is divided into 15 groups, each with a different focus, and each group must develop policy recommendations and legal language for Boston by the end of the semester. My group focuses on how Boston can use its curbs and streets to ensure that autonomous vehicles are adopted in a way that furthers Boston’s stated goals of decreasing climate emissions and making its transportation system more equitable. Over the course of the semester, the class engaged with weekly readings, videos, and guest lecturers, culminating in short responses. My eight responses are compiled below, along with a longer reflection.

1. What are Autonomous Vehicles?

Autonomous vehicle technology is under-deployed for narrow uses.

There are still serious engineering and marketing problems impeding widespread commercialization of autonomous vehicles. First, sensor costs are still rather high. TheHUB video posited that fully autonomous systems add about $10,000 to a car. Kelly Blue Book estimates that the average price of a new car in 2018 is $36,270. Adding an additional 28% in costs onto the average retail price of cars is a hefty upcharge. Moreover, the average American car is almost 12 years old, which hints at the unaffordability of new cars for many people at today’s prices. Admittedly, there are financing models (service fleets) and technological progress (cheaper and better sensors) that can rectify this issue. Emilio Frazzoli of NuTonomy does not think sensors are a major adoption barrier. Second, high definition maps and ground-based systems that augment GPS are not widely distributed. By way of comparison, a few major American airports are just now experimenting with complicated and expensive ground-based GPS enhancers that significantly increase the accuracy of the GPS output. Without this system in place, commercial airliners must land in inclement weather using ground-based radio signals. Autonomous vehicles need similar map granularity and/or augmentation systems wherever they operate; not just down to the space of a single runway. Yet, again, Emilio Frazzoli thinks that HD maps will become ubiquitous. Third, Sertac Karaman raised some yet unsolved programming issues, such as how to sequence vehicles at high-traffic intersections. Other programming issues revolve around the use of cheaper sensors, like cameras, which solve the price issue raised above, but are then harder to work with in inclement conditions.

Even with these issues, there are still a range of operations that autonomous vehicles can undertake. JFK and SFO, two of America’s busiest airports, currently run autonomous Airtrains. Similarly, the Honolulu Rail Transit Project is an operational autonomous train. The functionality of these systems is facilitated by what Chris Gerdes called the “operational design domain.” As he said, autonomous systems are easier to deploy when the operational parameters are narrow. Autonomous trains do not interact with other vehicles, run along predetermined routes, and stop at specified locations. Autonomous vehicles as popularly conceptualized can also be deployed today for similar narrow uses, with the benefit that their technology and design is such that their operational domain can expand as the software improves, uses are tested elsewhere, society accepts their broader integration, and infrastructure adapts. For example, the MBTA SL1 and SL2 lines might be made autonomous. These buses encounter more uncertainties than autonomous trains, but their routes are still narrowly defined, their lanes are often separated from other traffic, and they also stop at predetermined locations. As the capabilities of autonomous buses are proven, they might then be expanded to other lines. Unfortunately, deployment in even these narrow cases is impeded by technological path-dependency and/or slow fleet turnover, workforce resistance, tepid public response, and safety concerns.

2. What Legal Powers Does Boston Have?

Legal authority to install infrastructure-to-vehicle (I2V) and V2I communication technology is split between Massachusetts and Boston. However, layered on top of that authority is a Federal ‘command-and-control’ framework for spectrum that limits innovation.

In some domains, where power resides is relatively clear. M.G.L.A. 85 § 2 provides that the Massachusetts department of highways “shall erect and maintain on state highways … and on all main highways between cities and towns … mechanical traffic signal systems, [and] traffic devices ….” Similarly, Boston’s Transportation Department has the “power to make, erect, and maintain … traffic signs, signals, markings and other devices for the control of street traffic in the city ….” Therefore, the State has authority over highways while the City has authority within its boundaries, excluding the highways.

There are some spaces where legal authority is not so clear. Specifically, the Massachusetts Bay Transportation Authority (MBTA) is a cooperative State-City transportation authority. Deliberations must be in consultation with local officials, since the MBTA Advisory Board, which represents cities and towns in MBTA’s service area, has certain veto powers. Still, decisions are often made by Massachusetts officials. Most members of the MBTA Fiscal and Management Control Board are also members of Massachusetts Department of Transportation (MassDOT), and MBTA’s General Manager is also MassDOT’s Rail and Transit Administrator. When the MBTA runs Bus Rapid Transit (BRT), such as the recent Roslindale pilot, most of the infrastructure it uses is probably Bostonian. To the extent the bus runs on City property, it uses City infrastructure over which Boston has authority. But if BRT requires barrier separation from other traffic within Boston, does that infrastructure development fall within City jurisdiction or MBTA jurisdiction? And if it falls under MBTA jurisdiction as far as power to build and maintain, will MBTA utilize City standards to the extent they differ from State standards? Also, will MBTA be willing to invest in vehicles with V2X/X2V capabilities alongside City investment in such infrastructure technology?

City standards may differ from Massachusetts’ uniform traffic sign standards. These standards “conform to” Federal highway standards and are found in the MassDOT Standard Sign Book. In fact, Federal regulations require that the Sign Book maintain “substantial conformance” with Federal standards. 23 C.F.R. § 655.603(b). Yet, despite the Federal conformance mandate, the Sign Book reads more like a compilation of “sign faces in use throughout the Commonwealth,” which “may be used,” by cities and towns as opposed to a mandatory set of rules that aim to achieve statewide uniformity.

More importantly for present purposes, Federal and State sign standards seem only relate to physical human-interactive aspects of signage. The standards are specific as to aspects of sign dimensions, colors, fonts, and reflectivity. This leaves a major gap in regulation of installation of communications technology. But what, at first, seems like a gap that might beget opportunity for Boston to creatively install leading edge technology, is actually heavily regulated by the Federal government through spectrum allocation.

I have reached the suggested limit of our blog posts, so I will leave the details of Federal spectrum regulation to someone else. To sum up, Boston enjoys substantial authority to install traffic management technology within its boundaries. A couple big challenges, assuming buy-in to such innovation from City officials, are coordination with MBTA and navigating the constraints of Federal spectrum allocation.

3. What Are State and Federal Government Actors Doing?

The American Vision for Safer Transportation through Advancement of Revolutionary Technologies (AV START) Act establishes “a committee to make policy recommendations to Congress regarding HAV data ownership, control, and access,” to improve vehicle safety. The committee can look to the National Aeronautics and Space Administration’s (NASA) Aviation Safety Reporting System (ASRS) as an example of a well-functioning incident reporting mechanism that contributes to improved transportation safety.

ASRS has two objectives:

            1) Improve the national airspace system (NAS); and

            2) Enhance the database for human factors research and recommendations within the NAS.

ASRS achieves these objectives through voluntary, confidential, and non-punitive reporting. Anybody involved in aircraft operations may voluntarily submit incident reports to NASA about any unsafe occurrence or hazardous situation. After review by ASRS staffers, the reports are de-identified by removing the report’s “ID Strip,” which contains the reporter’s personal information. Once the incident information is coded into the ASRS Database, both the physical and electronic copies of the original report is destroyed to ensure confidentiality. Finally, the Federal Aviation Administration (FAA) will not, in most cases, levy a civil penalty nor suspend an aviator’s certificate if that person filed a report with ASRS within 10 days of the incident.

In place since 1976, ASRS has a long and successful record. In 2017, the program received over 94,000 reports, over half from air carrier pilots. Those reports helped improve the NAS. Some recent examples of system improvement include clarifying airport maps, enhancing airplane manuals, and renaming navigation reference points with confusing names.

This type of incident reporting system can be adapted to autonomous vehicles.

A reporting system can be implemented when drivers and autonomous technologies co-exist. The purposes of such a system would mirror those of the ASRS: to improve system safety and enhance human factors research. The fact that most ASRS reports are submitted by professional pilots suggests that any autonomous vehicle reporting should also focus on incentivizing professional drivers like long-haul truckers, van drivers, and taxi drivers. And the safety improvements might mirror those from the ASRS: updating maps, learning about vehicle component malfunctions, or redesigning traffic signs.

A reporting system can also be implemented independently of drivers for completely autonomous vehicles. Vehicles might periodically push sample data to a federal database or report any activity outside specific parameters. Again, the objectives and improvements might mirror the ASRS.

Many programmatic ASRS elements must be reconceived for autonomous vehicle implementation. Is the Department of Transportation the best agency to manage an incident database? Will states provide qualified immunity to vehicle operators, whether they be drivers, manufacturers, or fleet owners, for most or certain road infractions, incidents, or accidents if reports are filed? What information should be collected, how much flexibility should reporters be allowed to provide data, and in what format should the data be provided? These are just a few of the questions that the AV START committee will grapple with, and reviewing the ASRS is a good place for the committee to start.

4. The TNC Story in Boston and Elsewhere: What Facts Do We Have?

The facts of TNC usage are leading cities to reallocate curb space. Cities should continue to actively experiment with curb usage as transportation dynamics change.

Several of the pieces mention similar TNC usage patterns. TNCs experience high usage outside of peak travel times, between 8 PM and 6 AM. TNC adoption is higher among younger people, especially the 25-35 age range. TNC usage increases with disposable income, the minimum threshold appears to be about $50,000 with even higher usage at higher incomes.

Several pieces also cite similar reasons that people use TNCs. The two that specifically interest me are (1) to avoid drinking and driving and (2) when public transportation is not consistent or is closed.

These facts are not lost on cities. Washington D.C. recently decided to redesign some of its curb space near Dupont Circle. Dupont Circle is a hub of nightlife activity, with official bar and club capacity for about 17,000 people on any given night. Started back on August 30, and continuing for the next year, curbs that used to be 60 free parking spots in the area on Thursday, Friday, and Saturday nights will be dedicated pick-up and drop-off space for TNCs from 10 PM to 7 AM.

Boston might adopt similar policy for similar areas. Faneuil Hall is a well-known nightlife hub. The area where North Street connects Congress and Cross Streets is surrounded by bars and restaurants. “Walking” along North Street in Google Street View, signs inform drivers that there is “No stopping 7am–6pm. Except Sat & Sun.” and that the area is a “Tow Zone. Cab Stand 6pm–7am.” This particular street, then, might be too lenient for cabs, which can clog up streets during busy times like 6:30 pm and also too lenient towards non-hire vehicles that might stop at the curb on Saturday night, blocking out taxis and TNCs. This street and others like it are ripe candidates for experimentation.

The general idea is simple and tested – reallocate curbside parking to dedicated and necessary use to reduce street congestion. Manhattan is a great example of curb allocation in action. One curb section might change from parking, to empty (for cleaning), to loading/unloading, before beginning its cycle anew. The policy trick is deciding what curb use is most valuable.

The two reasons that people cite for using TNCs can guide policymakers’ focus in looking for areas ripe for curb reallocation. Areas with a high concentration of bars and restaurants. Areas with a high concentration of professional services or others who work long or odd hours. Areas around public transportation itself, even when that transportation is not running. In tandem with following citizens’ desires, cities can and should rely on facts when making these allocation decisions. There is a wealth of data on pick-up and drop-off locations and times. One example is the publicly available New York City taxi data. Combined, understanding peoples’ preferences and usage data are powerful enablers to make streets more responsive, dynamic, and efficient.

5. What Levers Does a City Have? What is Singapore Doing?

Several cities are integrating transportation network companies (TNCs) and autonomous vehicles into their public transportation networks. Phoenix is working with Waymo to pilot first mile-last mile service for certain customers. Gothenburg, Sweden launched an autonomous bus. And Charlotte, North Carolina has partnered with Lyft in an attempt to make that City’s public transportation more flexible, dynamic, and responsive. Integrating autonomous vehicles into public transportation networks provides cities a hook to exercise potentially significant of leverage over TNCs and autonomous vehicle operators: the request for proposals (RFPs).

RFPs are documents that announce what the publishing entity wants to buy. Public buyers are subject to various rules about how to promulgate RFPs and review sellers’ bids to pick winners. Without getting too hung up on the details, public procurement law in Europe (similar principles apply in the US, but I am only familiar with Europe’s rules) regulates process, transparency, and discrimination, but leaves substance largely to the public agency.

Public RFPs are also big business. According to the OECD, public spending in the United States fluctuated between 25-30% of total gross domestic product for at least the past 10 years. While the Federal annual budget is close to a whopping $4 trillion, states spend about $1.3 trillion annually and localities add another $1.5 trillion per year. Of that, about $300 billion flows into transportation. There is a lot of money to be made providing services, and transportation services, at governments’ behest.

Substantive procurement discretion couples with funding to provide governments with leverage that they can use to learn more about autonomous vehicles, control how autonomous vehicles are deployed, and more. Governments’ leverage is enhanced by the fact that autonomous vehicle operators seem aware that they need government cooperation to realize the potential of this nascent technology, so winning cooperative projects can separate market winners from pretenders.

The potential power of public procurement can be seen in Singapore’s 122 page Request for Information. Singapore asks for more than just data – it asks for a comprehensive plan and vision, which will help its decision-makers understand how well bidders’ dreams mesh with their own. Singapore also poses many questions about privacy, maintenance, and pricing, which will elicit numerous ideas and help Singaporean authorities to engage creatively with new modes of implementation. Singapore seems to be using its RFI quite effectively – other cities can do the same.

6. Labor Issues (in NYC)

This week’s articles paint a rather bleak picture. Drivers are psychologically manipulated by companies that enjoy informational and organizational advantages. The legal relationship is hollowed from a regulated employment contract to a mere use contract. And, in response, activists beat the old war drums for supply limitations or minimum wages. It is no wonder that Mr. Ochisor was “mad at the politicians,” and “mad at Silicon Valley.” The status of TNC drivers presents an opportune example to explore our assumptions about the legality of employment.

Employment contracts are the legalized rental of humans. In England you can “hire” a car much like Cravath might “hire” an HLS graduate. The idea of employment as human rental was openly pronounced in the canonical economics textbook Economics, by Paul Samuelson.

Modern human rental reflects old ideology about consent. One line of justification for monarchical rule rested on the consent of the people. However, that consent did not limit the monarch to representing his subjects’ will – it was not a delegation from “we the people” to the monarch – but was a complete alienation of power and decision-making. This same consent-as-alienation idea also justified voluntary slave and coverture marriage contracts. We now largely agree that consent-as-alienation is an unacceptable intellectual and legal structure for the examples just cited. Yet corporate workers are subject to the same loss of power in the employment context that the people, slaves, and wives were in those other contexts.

Corporations are major governance institutions. They make a range of vitally important decisions from building factories, to investing in employees’ retirement, to lobbying for or against technical regulations, and on down the line. And corporations are not democracies. Some of America’s largest public companies, like Facebook, are controlled through dual-class voting shares by their founders. Even in other public companies, and Uber is considering going public at a $120 billion valuation, decision-making is largely left to executive management, sometimes with input from billionaire investors. At no point in this story do ‘retail investors’ or rented humans have any meaningful input.

In the context of Uber, recognizing that drivers give up their autonomy interests through rental contracts that alienate their decision-making power to Dara Khosrowshahi makes the legal battle about classifying them as employees or independent contractors a bit quaint. Sure, drivers-as-employees have more protections in their legal rental relationship than they do as contractors. In either case, however, the drivers barely make enough to survive, much less capture the full value of their work, and are disenfranchised from influencing decisions about a subject which they know better than almost anyone else.

The issue is two-fold. The first is democratic. The democratic deficit in private sector governance is a problematic flaw in our society. The second is economic. Income inequality is much smaller than wealth inequality, and ownership, not wages, beget wealth. Human rental excludes workers from both.

7. Los Angeles’ Public Transportation Department

This week’s materials, especially the conversation with Scott Corwin and Seleta Reynolds, induced me to reexamine the nature of public transportation agencies. In particular, Corwin and Reynolds disagreed about whether public agencies have the capacity to create a viable and popular MaaS platform that integrates all modes of transportation around a metropolitan area. Reynolds thinks it is preferable over public-private partnerships, while Corwin thinks private developers are necessary. While Corwin’s position strikes me as more probable, I think cities and their transportation departments should reimagine their capacity to act as MaaS providers.

The typical transportation model inserts public transportation as an alternative system to private transportation. Sometimes the two are in competition; sometimes they are complementary. Each runs different modes (trains versus cars) and capacities (bus versus small car). In many ways, both public and private transportation fill gaps left by the other system. In the traditional model, both private and public own and operate transportation assets to meet needs unfulfilled by the other system.

City agencies could reimagine their role from that of an asset-heavy transportation system operator to one of an asset-light transportation coordinator, much like TNCs but on a much broader, multi-modal scale. What would this entail? First, a city must have robust internal software development and data privacy teams. A desirable MaaS application will take an enormous amount of back- and front-end development. Second, the city could progressively sell its transportation assets to private operators, perhaps with some contractual covenants that the assets, or better yet their seating equivalents, will not be retired from the metro transportation system for at least a certain amount of time. Third, that city would need to use monetary incentives to induce private actors to service underserved areas. Those incentives might be allocated through credit given directly to low-income households. Fourth, the asset-light transportation department should also use monetary incentives to encourage the use and development of preferred modes. Just because a city steps away from owning the subway does not mean that it should not continue to encourage subway ridership.

An asset-light public transportation model has several redeeming characteristics. It makes public transportation agencies much more flexible. They might allocate substantial funds to subsidize subway ridership today and shift those funds to shared autonomous vehicles later. Relatedly, cities could be more responsive to citizens by changing transportation budgets as needs arise rather than suffering the path-dependency induced by asset purchases. Also, cities would have much better data about when and how their population travels, enabling much smarter allocation of transportation subsidies to serve the underserved and enhance efficiencies.

That data is critical to developing a core asset that I don’t think cities can privatize: transportation infrastructure. No matter how asset-light cities get they still need to build roads. The questions that cities need to answer is how and where to build to support the transportation dynamics of their city. Shifting public investment from operational asset investment to platform provision and systemic nudging would provide unprecedented insight and dynamic influence.

8. Electric Vehicles, Autonomous Vehicles, and Regional Energy Issues

When I worked for Ford Motor Company I drove a Fusion plug-in hybrid. Unfortunately, I lived in a big old multi-building, multi-unit complex that did not have EV parking, so the all-electric range went virtually unused. My story is an example of why initiatives like Vermont’s EV Readiness Building Codes are important to support the adoption of EVs.

Vermont’s program requires multifamily developments with more than 10 units to make at least 4% of parking spaces EV-ready, as well as establishing requirements for new commercial buildings. Establishing EV requirements for new buildings is a relatively efficient regulation. It is cheaper to install wiring and infrastructure during site construction than to retrofit existing sites. Also, electrifying new parking will match the roll-out of EV-enabled parking with actual vehicle adoption.

However, the impact of EV-ready parking in multifamily housing should not be overstated. People who rent tend to be younger, poorer, and more urban. Therefore, they are also less likely to own, lease, or use cars, especially new cars. That said, those who live in buildings without EV-enabled parking have no choice about whether to join the growing EV market, and there are surely many who would join the EV market, if given the infrastructure.

It seems like there are at least two things states can do to enhance the value of their EV-ready building codes.

First, states can enhance the value of their EV-ready building codes by digitizing the location and type of charging available when those chargers are installed. In this way, cities will have a database ready to support a future of ubiquitous electric fleets (or any other future in which it may be useful). It is possible that those fleets will use centralized charging, but they may also need access to some distributed charging and the pre-installed capacity can serve that need. The UC Davis policy brief frets that charging installed for personal-use vehicles may not be well-suited for autonomous fleets. I agree, but we do not know yet whether that is true or not. By the time autonomous fleets scale, it is quite likely that charging infrastructure will go through at least one generation of development, allowing for a rethink and redeployment of charging facilities, if they are actually very badly deployed for fleet operations. If states digitize the EV-charging network, they should also require owners to periodically update the state on the status of those chargers, so uninstalled or inoperable stations do not remain in the digital network.

Second, they might push for EV-readiness in new developments and homes, which would help enable richer, older, and more rural populations. They could require developers to include fast-chargers in a certain percentage of new homes or at every block. Or they might engage in information and marketing campaigns to encourage individual homeowners who are remodeling or rebuilding to install advanced EV charging. Even if the homeowner does not use the infrastructure, it might increase the home’s resale value.

9. Reflective Essay

I enrolled in this class to stay updated on the development of autonomous vehicles, understand better how they might change transportation, and think about policy interventions to encourage the augmentation of their benefits and mitigate their negative consequences. The lab serves all of those goals and more. Particularly, Group 3’s work on policy recommendations for Boston’s curbs and streets is interacting unexpectedly with the views that I am forming around autonomous vehicle development, and shifting my focus when thinking about the lab. More concretely, I am skeptical about the timeline for the roll-out of fully autonomous vehicles with car-equivalent operational freedom. This skepticism feeds directly into how I approach developing policy recommendations for Boston’s built environment, targeted towards achieving Boston’s 2030 goals and in the context of the generational battle to redesign society towards net zero emissions.

Fully autonomous and operationally unconstrained vehicles hold enormous transformational potential. I love dreaming about the death of personal car ownership, eradication of unused parking, a dynamic and efficient transportation network with near-perfect safety, and distributed batteries that not only power the vehicles but also store energy from intermittent renewable sources. But these dreams cannot be implemented today because they do not align with the current state of autonomous vehicle, much less battery, technology.

The lab’s opening materials on technological substance were very informative in this regard. John Leonard was especially impactful. His personal experience with autonomous vehicles of all kinds gave credibility to his skepticism about the timeline for the development of the technologies that underlie autonomous road transport. That lecture dovetailed with Emilio Frazzoli’s lecture demonstrating autonomous vehicle capabilities from many years ago, as well as NuTonomy representatives discussing the truly circumscribed, narrow nature of their operational mandate. This information conformed to my prior experience.

I came to the lab with a notion that autonomous vehicle commercialization might be slower than advertised. That said, my perception of the commercialization timeline became compressed the more time I spent away from autonomous vehicle engineers. In fact, in my first post I struck an optimistic tone, claiming that autonomous vehicles might be used today in circumscribed operational environments like bus rapid transit. However, hearing from engineers has re-implanted my commercialization incrementalism. For example, a guest speaker (I believe it was a NuTonomy employee) mentioned the radically different physics of Boston buses compared with small personal cars, and the difficulty of translating NuTonomy’s experience to such a different vehicle. Moreover, NuTonomy’s outlook on how the business might develop is a bit pessimistic, especially for a city like Boston. That is, autonomous vehicle companies will monopolize particular metro markets because generalizing to different cities is too hard.

This view of the commercialization of autonomous vehicle technology as a slower incremental process has a direct impact on my group work. Put bluntly, when thinking about changes Boston might make to curbs and streets in the short-term, and even in the medium- or long-term, in many instances it is better to ignore the possibility for autonomous vehicle deployment. Several factors are especially relevant to this conclusion.

First, Boston has particular transportation goals. There are myriad goals listed in the Go Boston 2030 report. Listing a few suffices to make the point. Boston aims to reduce regional vehicle miles traveled by 5.5% below 2005 levels by 2020. The city wants to reduce greenhouse gas emissions from transportation by 50% of 2005 levels by 2030. And Boston wants to reduce the transportation cost burden for very low income citizens from 33% of income to 13%.

Second, Boston has particular emissions goals. One was already mentioned: 50% reduction in transportation emissions below 2005 levels by 2030. Boston’s environmental goals are not the isolated environmentalist concerns of a liberal coastal city. They interact with international action, emanating from the Paris Climate Agreement, and the evolving science and experience about how we need to change our society to both mitigate further harm and also adapt to a new climate reality. Go Boston 2030 acknowledges that 25% of Boston greenhouse gas emissions are from transportation sources.

Third, nobody knows exactly how autonomous vehicle technology will be commercialized. Many companies envision a world where autonomous Uber fleets roam city streets and serve most transportation needs. But so far, that is not the world we see. Driver assist technologies, close cousins of autonomous vehicle technologies, are today commercialized through the sale of private vehicles. As discussed above, autonomous vehicles might be commercialized through proof-of-concept single-type-operations like particular bus routes, subject to regulatory approval or simply out of an abundance of caution. Even in a ubiquitous fleet world, private car ownership will not disappear overnight. The average car is well over 10 years old, and so does not have the embedded technology to be easily upgraded into an autonomous actor, and it will take some time for cars to age out of use.

Fourth, nobody knows exactly how autonomous vehicle commercialization will interact with other types of technology. Autonomous vehicles as tested by Waymo or Uber may turn out to be the best way, or even just a socially serviceable way, to store and use intermittent renewable energy. But perhaps autonomous vehicles and battery technology are better paired in bicycle or scooter-like packages. Along those lines, competing modes of transport might become unexpectedly favored over those we see tested today, such as the rise of powered scooters.

Fifth, Boston must exert control over the means to achieve its transportation and emissions goals, in the context of uncertain autonomous vehicle commercialization, to design the type of city that Boston leaders envision. The timelines to achieve those goals do not necessarily align with the timeline for autonomous vehicle rollout. When John Leonard polled the class asking when we think autonomous vehicles will reach commercialization, I picked the farthest date out, which was well past 2030. Boston leaders, planners, officials, and citizens do not have that much time to design and implement the city they want to live in, either politically or environmentally. Politically, Boston officials must take action tomorrow and in the next few years to improve reliability, and increase access and safety, with the technology and options at their disposal today. Environmentally, Boston officials cannot count on autonomous vehicles to be a panacea that brings emissions to zero. On both fronts, City officials cannot abdicate their responsibility to actively design a built environment that encourages zero- and low-emissions transportation in the reality that exists simply because there is an uncertain utopian technological revolution on the horizon.

Most importantly in this regard, autonomous vehicle developers will work within the physical and legal parameters set by Boston. A NuTonomy representative said that the company is building its vehicles to operate independently from the urban environment. That is, the company does not want to rely on city infrastructure to guarantee the operational reliability of its vehicles. I imagine that all autonomous vehicle developers hold some semblance of a similar attitude. Imbedded in this desire for operational independence is an implicit recognition that transportation infrastructure is in many ways a neutral platform. Roads are generally flat and paved, which makes them open to new technological modes and, if new technology becomes sufficiently important, the applicable rules-of-the-road can be changed fairly easily.

Therefore, Boston should plow ahead and redesign itself to meet its goals in a way that does not rely on autonomous vehicles. In most cases, policy interventions are neutral towards autonomous vehicle adoption. For example, reassigning parking spots as pick-up and drop-off locations is neutral towards the development of autonomous vehicles. More controversially, encouraging curbside shared-bike access to increase the catchment areas for trains and buses is autonomous vehicle neutral. Autonomous vehicles might be implemented in narrow operational domains like trains before they ever become ubiquitous, in which case such policy supports autonomous vehicle adoption. However, autonomous vehicles might commercialize in the same role as the bikes, serving as the first mile/last mile link, in which case the policy works at cross-purposes to autonomous vehicle adoption. Critically, nobody knows when nor how autonomous vehicles will roll out, nor their environmental and social valences. Meanwhile, Boston has promulgated specific and clear goals outlining the type of city it wants to become.

So, in sum, the transition of my perspective during the semester so far has been drastic, facilitated by our material on the technology and Group 3’s focus on curbs and streets. I enrolled in this lab to think about the impacts that autonomous vehicles might have on society. However, given my incrementalist view about autonomous vehicle adoption and its uncertainty, combined with its interaction with the timelines for political and environmental action, I am now more conscious of the limitations of autonomous vehicles with respect to solving, in part or in whole, transportation and environmental problems. Working with this context in mind, my approach to thinking about policy recommendations for Boston is, as an initial matter, to ignore the possibility of autonomous vehicles, and to simply think about how officials might further their goals in the world we face today. Only then, once a recommendation is formulated, do I even try to balance the interaction of that policy in today’s world with how it might interact with possible autonomous vehicle futures.