Self-Driving Vehicles Conference 2015

+ Jonathan Stewart, Alain Bertaud

Since the car was invented and spread to the masses in the early 20th century, its concept has remained essentially unchanged. Cars have become faster and safer, but in developed countries they are still mostly the 1-driver, 3-passenger design that came off of the production lines one hundred years ago. The advent of self-driving vehicles (SDVs), however, presents the developed world with the first prospect of a transportation revolution in a century.

This revolution will have far-reaching implications for our economy and society, and its impact on our cities will be fundamental to how it impacts our lives. To explore these effects, the NYU Marron Institute of Urban Management -- with support from Google -- hosted a conference on self-driving vehicles on May 28 & 29, 2015. The conference focused on self-driving vehicles’ impact on urban areas’ labor markets, transit, and land use. About 25 academics attended, several of whom gave presentations and wrote papers, which can be viewed on the conference website.

Several additional themes emerged from the presentations and attendee discussions, including:

  1. The prospect of cities’ entire transportation system consisting of public transit, with cars progressively disappearing (mostly a European issue)
  2. Simulation of potential impact of SDV in specific cities or states
  3. Potential of self-driving cars when combined with trip sharing (Uber, Uber Pool type)
  4. Technology and automation in transport: prospects and limits
  5. The need for more precise terminology when dealing with SDVs and transportation issues

Most of the papers assume that self-driving vehicles will constitute 40% to 100% of vehicular transportation. No paper addressed the initial stage of self-driving vehicle adoption where benefits accrue to individual users but do not yet generate any positive externalities on congestion, transport time or parking space.

Many European planners and mayors believe that cars should and will disappear as a means of urban transport. Felipe Calderón and Al Gore gave voice to this view at the World Economic Forum in Davos in 2015. If this prediction is true, there is no reason to develop self-driving vehicles. However, Rémy Prud’homme, the only European speaker at the conference, debunked this notion in his paper and presentation. It remains an active debate in Europe.

In "Self-Driving Cars and the Efficiency of Cities," Rémy Prud'homme assumes that all cars, or most of them, will become self-driving. SDVs will replace individual cars and unshared taxis but will not be part of a shared fleet. He documents:

  • Safety gains
  • Comfort gains
  • Increased speed gains
  • Increased labor market gains
  • The economic gain from having a generalized use of SDVs will represent about 8% to 10% of the GDP of a large city such as Paris. The biggest gain is from the comfort gains and the increased labor market efficiency, rather than from the safety gains or the increased speed gains
  • This is a very different conclusion from the Beijing paper. The labor gains are also seen differently in Paris and in Beijing. In Paris, the gains come from larger labor markets. In Beijing, from reassigning workers to more productive activities

In “Perspective View of Driverless Cars in Beijing,” Guo Jifu evaluated and quantified the potential impact of self-driving vehicles on urban transport in Beijing, including:

  • Time saved by using self-driving vehicles as a complement to public transport, as door-to-station and station-to-door trips
  • Using self-driving vehicles for short distance trips
  • Decreasing parking requirements due to self-driving car usage
  • Increasing labor productivity by releasing drivers (estimated at 320K individuals in Beijing) into more productive economic activity

The increase in labor productivity due to self-driving vehicles is an interesting perspective on an important issue, as most policy-makers and citizens are concerned that self-driving vehicles will lead to a net loss of low skill jobs. While this transition will generate winners and losers in the short run, the more long-run traditional “economist” view is that technological improvement will cause labor productivity to increase.

In "An Analysis of the Impact of Self Driving System on Transport & Land Use: A Case Study of Korea," Donghyung Yook of the Korea Research Institute for Human Settlements presents the only conference paper that explores the change in land use due to SDVs. To do this, he uses a simulation model that compares the spatial structure of Seoul MSA with and without self-driving vehicles. He concludes that self-driving vehicles (100% penetration) will lead to a more dispersed development of urban space in the MSA by opening much more land for development. Although the paper is not explicit about it, it can be inferred that the use of SDVs will significantly lower housing prices. The paper also tests and measures:

  • The impact with a progressive introduction of SDVs
  • Traffic safety enhancement
  • Travel time savings & congestion reduction

Dinesh Mohan of the Indian Institute of Technology wrote "Autonomous Vehicles and Their Future in Low and Middle-Income Countries." He sees few advantages of SDVs in India and therefore a very slow introduction there. His main points were that:

  • The substitution of expensive technology for cheap labor is unlikely to happen. A chauffeur-driven car offers more advantages than SDVs for people who can afford them
  • There could be some application of the technology applied to public transport, not as a feeder to stations, but on public transport vehicles themselves for routing purposes
  • The chaotic traffic situation of Indian roads and the lack of road markings would make it difficult to introduce SDVs there. Most drivers will soon learn how to “game” SDVs in a confused traffic situation with few rules

In "Congestion, Economic Performance, and Autonomous Vehicles," Clifford Winston analyzes the benefit of reduced congestion to travelers and to the non-transport part of the economy. This is the only conference paper that evaluates the impact of SDVs on non-transport economy. He finds that:

  • Benefits depend on market penetration—50% penetration could reduce congestion delays 50% and could yield annual benefits to US travelers of some $200 billion
  • Self-driving vehicles have the potential to greatly improve infrastructure efficiency, which would generate large benefits to travelers and non-transport sectors of the economy

In "Trip Sharing in the Era of Self-Driving Cars," Michael Szell of Northeastern University considers the potential of SDVs when combined with trip sharing. He finds that:

  • The cumulative trip length of all taxis in the system can be cut by 40% if passengers are willing to share a cab. If taxis become SDVs then user costs are lower and taxis will be more frequently used
  • The lines between private and public transportation and between individual and mass transportation will be blurred
  • Self-driving technology multiplies the environmental benefits of trip sharing

Finally, a recurring theme of the conference was the need to develop a common lexicon to use when discussing self-driving vehicles. Most participants disliked the term driverless car, both because the vehicle has a driver (the computer) and because the term “car” is too strongly associated with the 20th century manifestation of a 4-person vehicle. Self-driving or autonomous vehicles may be two- or three-wheeled individual enclosed vehicles, a couch in a box with no steering wheel, or four seats around a table on wheels. Google uses the term self-driving car, while the Marron Institute has opted for self-driving vehicle.

“Point-to-point” transport accurately describes the kind of transportation that cars provide, while transit generally indicates public mass transportation links, like trains or subways, which provide “station-to-station” transportation. “Feeder” may be too narrow to specify the kind of scattered links to mass transportation that will be necessary in increasingly disperse cities. “Point-to-hub” may work better at signaling the kind of critical linkages to mass transportation that self-driving vehicles may be singularly capable of providing. Currently, a common trip would then consist of a “point-to-station” (walking to a bus stop). Followed by a “station-to-hub”(a bus trip to a hub metro station), followed by a “hub-to-hub” (the metro ride itself) then a “hub-to-point” walking to final destination.

Furthermore, several terms seem to need revision or additional specification. “Sharing” may be an insufficiently precise term, as it can indicate one vehicle serving two travelers at the same time or one vehicle serving two travelers in succession. “Congestion” is also problematic, as it can point to a capacity constraint (as on the subway) or gridlock that actually reduces the system throughput (as in overly crowded roadways).

Conference working papers and presentations may be found on the conference website. In addition, Paul Romer, Director of the NYU Marron Institute, had some thoughts on the conference, and the terminology in particular, that may be found here.

Special thanks to all conference participants and organizers: Solly Angel, Alain Bertaud, Marie-Agnes Bertaud, Jit Bajpai, Alex Blei, T. Donna Chen, Jeremy Coleman, Brandon Fuller, Eric Jaffe, Guo Jifu, Jennifer Haroon, Pilar Harris, Sarah Hunter, Kim Kyung-Hwan, Zoe Johnson, Sarah Kaufman, David King, David Levinson, Kelly Miller, Dinesh Mohan, Kaan Özbay, Rémy Prud’homme, Paul Romer, Jonathan Stewart, Arun Sundararajan, Michael Szell, Anthony Townsend, Clifford Winston, Dong-Hyung Yook

Photo courtesy of Ed and Eddie.

Back to top
see comments ()