The 10 Assumptions of Autonomous Vehicles

Planning for Our Future Metropolises

Yes, there’s a discussion [about incorporating autonomous vehicles into planning efforts at our MPO]. We don’t know what the hell to do about it. It’s like pondering the imponderable. — Planning for Cars That Drive Themselves

There’s something comforting about urban planners (especially transportation planners) acknowledging a great unknown, and simultaneously, not being deterministic about what the future should look like. Even it amounts to a ¯\_(ツ)_/¯, it’s a welcome change for the field considering how stubbornly slow it’s been in recognizing the ongoing changes in mobility patterns:

A Series of Wildly Inaccurate Projections from US DOT Transportation Planners — Frontier Group

Of course, anticipating future impacts and needs is the essential role of planners, and yet, as of the end of 2015, less than 10% of regional transportation plans even consider the impacts of driverless technology. Uber’s unleashing of self-driving taxis on the streets of Pittsburgh next month should shake any complacency in city and regional officials. As Robin Chase said, this is going to happen. An autonomous vehicle (aka self-driving car, aka driverless car) future approaches.

In the past months, scores of articles have been written (by people outside of the planning world) on the latest driverless tech developments, and how our lives will be changing. One thing’s certain: people are excited about autonomous vehicles (AVs). Undoubtedly, we are in for a great shift in mobility, perhaps the most disruptive of many of our lifetimes. Other disruptive, and no less exciting, technological trends are on the horizon — smart cities/IoT, augmented/virtual reality, drones, large-scale 3D printing, etc — and, depending on the month, these may get more media coverage than AVs. However, AVs’ societal impact will be grander, changing inherent aspects of how most of us move around in the world. Similar to how cars shaped the American landscape in the 20th century, AVs have the potential to reconfigure our relationships with the land and built environments around us, challenging our fundamental conceptions of access, distance, land use and allocation.

In the face of these uncertain impacts, the potential for a significantly worse status quo is worrisome. Again, stealing from Robin Chase, Self-Driving Cars Will Improve Our Cities. If They Don’t Ruin Them. Worst-case scenarios such as explosive sprawl, endless traffic jams, elimination of millions of jobs, or widespread hacker-induced crashes are perhaps dystopian enough that the powerful and political establishments would seek to remedy them. However, many less-than-optimal outcomes are possible and may slip through the cracks without pushback or advocacy, evidenced by incidents such as a politician insidiously pointing to an AV future as a reason to not invest in public transit today. Indeed, the name of the AV companies’ new lobbying group, Self-Driving Coalition for Safer Streets, smacks of astroturfing. Does a “safe street” include people? Or allow for alternative modes? Isn’t the safest street one that doesn’t allow cars at all? Letting these companies define our communities the same way that GM was able to at the World’s Fair is a mistake we can’t afford to repeat.

Uber, in particular, has been called out for a lack of concern for those being disrupted and envisioned as a potential provoker of a new streetcar conspiracy, precipitating the demise of public transit systems. Private regulatory capture is already a concern with technology companies innovating in the mobility space with little public input. Surge pricing is a case in point: despite being a de facto form of demand or congestion pricing, proceeds are captured by private firms (and, for now, drivers). A congestion pricing scheme collecting proceeds for public infrastructure improvements, while incentivizing shared mobility, is the more desirable scenario, and can still be implemented.

As the above demonstrates, the transition to AVs presents tremendous opportunities for improving our cities while correcting previous poor decisions. So my message is to urban planners: This is not the time to throw our collective hands in the air. AVs are coming (and soon!), and they will fundamentally change metropolitan regions and their planning (for better or worse).

Despite the dizzying potentials for disruptions, much of AVs’ impacts are in fact conceivable and must be clarified. To help those unsure of how or what to think about these implications, I’ve compiled assumptions about the future of AVs, with the goal of defining some boundaries on the vast possibilities.

This is not about how the transition to an AV future happens or why — the transition will be far from seamless — it is to highlight that new (undeniably exciting) technologies and developments may lead us down particular paths, and we might not like where they end. There are still more questions than answers, but I aim to cover some of those questions in a later post.

The Future of the Auto Industry

AV Assumptions

1. “Driving” with AVs will be significantly safer than with human-operated vehicles. Extending the concept of current crash prevention systems, safety is the most salient aspect of autonomous driving given that human error accounts for 90% of car crashes. Though accidents get all the press and to date AVs have only demonstrated modest improvements in safety (up to 50%), their connected, machine-learning foundation suggest this will only improve. For example, with X’s Self-Driving Car, the aim is to be at least ten times safer than current rates. Long-term, AVs will be able to avoid nearly all accidents (with some limitations).
2. Once demonstrated to be safe, the demand for AVs will be large and adoption swift. The promised safety is why the Obama administration has requested $4 billion to facilitate self-driving. Meanwhile, insurance companies are likely to encourage autonomous systems through lower premiums, as they do for other safety technologies. Of course, with autonomous vehicles, the necessity of independent auto insurance providers is called into question (or not). Already Volvo, Mercedes & Tesla have proclaimed that they will assume responsibility for accidents involving their AVs.
   For consumers (that is 90% of the population), the prospect of a relaxing or productive commute will be impossible to resist. The commute has been consistently rated as people’s worst part of the day, while its negative impacts are well-covered. Cost may be an impediment with some projecting self-driving systems adding $5,000 by 2030. However, there’s reason to think this is far too high — as Sertac Karaman at MIT has noted, AV’s $100 thousand computers in 2007 have been surpassed by $500 embedded computers in 2015, and the leading LiDAR manufacturer predicts a similar price trajectory for their sensors. Individuals and companies are already working on retrofits.
3. AVs will accelerate shifts in ownership models. It’s clear that the transition from the personal automobile to shared solutions is already underway, as described in the above image. What’s also clear is that the economics of shared mobility favors AVs even more than for personal vehicles. The massive cost savings found by operators from eliminating drivers are difficult to overstate. According to Uber’s CEO, “prices will fall so low that the per-mile cost of travel, even for long trips in rural areas, will be cheaper in a driverless Uber than in a private car.”
4. The adoption of shared autonomy will be uneven. Mobility aggregator apps are already integrating the existing, diverse existing car- and ride-sharing solutions. These will likely converge further such that a user within a metropolis will be able to select the best mode for a specific trip. Yet, there are upper bounds to shared mobility, despite claims of the eventuality of an ownerless society. Yes, in cities each shared vehicle can take upwards of 10 cars off the road, yet trip overlap (which enables sharing) drops precipitously in less-dense areas, and especially in rigidly hierarchical street networks — i.e. arterials and private, cul-de-sac developments. Though households in these contexts will be able to reduce the number of vehicles owned.
5. AVs will run on electricity. And AVs will be the key driver for the transition to electric vehicles (EVs). Ignoring for a moment the spectacle of Tesla Model 3 reservations, EV adoption has not matched expectations. This is despite EV’s acknowledged better driving experience. The two assumed impediments to widespread EV adoption are high cost and “range anxiety,” and both will be less relevant in the near term. EV prices are dropping, perhaps not as quickly as some would hope, but by 2022, battery electric vehicles are expected to be cheaper than the gasoline alternative.
   Wireless charging will enable AV recharging without human assistance, which will be a necessity if current non-drivers are allowed to operate AVs. AVs then potentially eliminate refueling as a concern for humans, and future wireless roads facilitate uninterrupted vehicle operation — perfect for ride-sharing services. Perhaps some heavy-duty vehicles will require the higher energy density of gasoline or hydrogen fuel cells, but overall the transportation sector will be greened significantly, in one estimate seeing a 90 percent decrease in greenhouse gas emissions and almost 100 percent reduction in oil consumption.
6. Autonomy won’t be limited to cars. Autonomous platforms — requiring only computing power, wireless connectivity, and sensors — can work for everything from tractor-trailers to personal scooters, to helicopters and drones, as well as vehicle formats that don’t yet exist. EVs have much simpler drivetrains compared to gas-powered vehicles, and, as demonstrated by Faraday Future’s proposed Variable Platform Architecture, can be right-sized to meet specific needs. AVs can be further shrunk by removing soon-unnecessary safety features (e.g. crumple zones) or purpose-built for commercial uses, including eliminating humans from designs entirely.
7. AVs will integrate into the Smart City. With the shift to constant connectivity, ubiquitous GPS and automakers boasting about their networking capabilities, the latest cars finally realize the vision of connected vehicles from the World’s Fair Futurama exhibit. Wireless standards are being designated for vehicle-to-vehicle and vehicle-to-infrastructure communication, while AV-makers develop enhanced street maps and traffic models. AVs will surpass current route guidance applications and use data from the Smart City to decide the optimal route, anticipating road work, coordinating with traffic lights, or avoiding roads with predicted slowdowns.
8. AV usage will be priced through some combination of VMT, congestion, and road pricing. The shift to EVs will cause the already precarious reliance on gas taxes to fail completely. Anticipating this, per-mile (VMT) fee experiments are ongoing, though with AVs, VMT will be measured seamlessly, in real-time. Optimal routing decisions will also be able to incorporate cost, not limited to VMT. Congestion or surge pricing scheme implementation will be simplified; with each vehicle on the roadway accounted for, individual vehicle’s contributions to congestion can be billed directly and immediately. Variable pricing for impacts to roadways (after all, weight impacts on roads exponentially), surcharges for unwanted traffic (e.g. in pedestrian areas or school zones), or new tolling mechanisms to fund roadway maintenance or new construction are all feasible.
9. AVs will substantially increase road efficiency, as well as roadway usage. The marginal cost of an AV mile is going to be negligible.
   The Six Efficiencies of Autonomous Vehicles:
   1. Elimination of slowdowns due to traffic incidents
   2. Closer distances between vehicle
   3. Narrower lanes
   4. Higher speeds
   5. Smaller or otherwise right-sized vehicles
   6. Widespread use of shared vehicles
   Those efficiency gains unfortunately don’t imply empty roadways. Eventually, the well established concept of induced demand — creating more space on the roadway, traditionally through road widenings, increases the number of vehicles and ultimately leading to more traffic — takes hold. Already, route guidance apps such as Waze are directing drivers to secondary roadways, improving the efficiency of the overall road network, and now leading to growing complaints about once quiet streets now filled with traffic. These efficiency gains pale in comparison to those six listed above, though given our history with roads, we can be certain that even these efficiency gains will eventually lead to more traffic.
   Congestion pricing or another type of discouragement will be needed to avoid a future of roads clogged with AVs. Electrically-powered and driverless, AVs’ operational costs will be minuscule. The economics will be similar to taxis — incentivized to drive as much as possible — even if the AV is owned rather than shared. Perhaps your AV can circle the block while you run an errand (avoiding parking), or even pick up a passenger and make you a buck. The lower stress and more pleasant experience with AVs’ will lower barriers to travel, allowing for longer trips, and more of them.
10. AVs will comprise a mobility platform. As low-cost, zero-emission, hyper-responsive, and ubiquitous mobility services are rolled out, conceptions of access and movement (of people, goods, and services) will shift. Those segments of the population currently unable to drive will have equal access to transportation options, goods producers will use one of many pre-optimized logistics providers for just-in-time or even predictive delivery, and the fledgling expanse of Uber for X companies will transition into a layer on top of the mobility platform. As the platform is standardized, secured, and optimized, new trip types will emerge of anything from transporting pets to off-leash parks, to on-the-fly electric grid rebalancing, to emergency mobile organ generation. Eventually this platform will grow to include autonomous drones.