Autonomous Vehicles: A Solution For...Which Problem, Exactly?
By Josh Linden
Buried within an extremely interesting debate in September at the Cato Institute between Jarrett Walker (Oregon-based transit consultant) and Randal O’Toole (Oregon-based anti-transit writer) is a brief discussion of autonomous vehicles. Will AVs be an “extinction-level event” for transit, as O’Toole claims, by providing residents with a more convenient way to get around in a far less congested road environment? Or will their overall impact depend upon the precise nature of their deployment (which may still be years away) and the specific problem they’re trying to solve, as Walker argues?
For O’Toole, the problem is relatively simple: (1) people love driving; (2) driving is often too difficult or takes too long; and (3) we should therefore make driving easier. The personal automobile was a transformative and liberating event for humanity, in his view, and national mode-share trends -- despite billions of dollars in transit subsidies to support and expand systems with falling ridership -- demonstrate an overwhelming and consistent preference for driving. O’Toole argues that we should therefore optimize our transportation system for people’s expressed preference -- i.e. make driving easier and more convenient in a way that can also generate new transportation trips (which then creates new economic activity as well). He believes that AVs can achieve this outcome by reducing energy inefficiency, congestion, and roadway crashes.
Walker flips this logic on its head, and tries to first disentangle the myriad of transportation challenges cities experience as a way to think more clearly about the specific role for new technologies (rather than start from the normative view that cars should determine how we plan for the future). For example, efficiency, congestion, and safety are all distinct problems that may require different solutions.
Can AVs address all of them? Well, the problem of energy efficiency could be addressed through electric vehicles. But what happens if those electric vehicles become AVs that run more often and therefore require more regular battery charging? Sustainability and efficiency at that stage depend upon the sourcing of the energy from the power grid -- and we are still, unfortunately, far away from 100% renewable energy at that level.
The problem of congestion is fundamentally a problem of space -- and urban environments are in many ways defined by limited space relative to the number of people (i.e. density). Can AVs optimize the use of this limited space in a way that significantly reduces congestion? That may depend upon another factor -- induced demand. If we make driving easier, then more people are likely to choose that mode. So while AVs may optimize space between vehicles and allow for somewhat faster (or more consistent) speeds, induced demand could easily lead to dramatic increases in the number of AVs on the road. Optimization of space is still constrained by the overall geometry, and more cars in the same roadway could offset the benefits of overall optimization. In these cases, large vehicles with multiple people could still be a solution (i.e. transit), and perhaps automated buses would be a more useful form of AV technology than cars -- particularly since they would follow fixed routes, which helps with optimization.
How about safety? According to data from the National Highway Traffic Safety Administration, nearly half of traffic-related fatalities occur in urban environments, with one-fifth of all fatalities on urban arterials alone. Since 90 percent of all motor vehicle crashes are caused by some form of human error, there could be an opportunity for AVs to make a dramatic difference in roadway fatalities. However, safety also means slower speeds to prevent sudden braking or swerving in response to unexpected conditions. What sort of unexpected things happen in streets and public spaces? Well...pedestrians, cyclists, scooterists, and others on active transportation will continue to use our city streets, and human behavior is unpredictable. AVs will have to account for the tremendous amount of street-level activity that cities generate, which often means slower, more careful driving, particularly within dense portions of a city’s grid network with shorter distances between intersections and high levels of mixed-use development. In these environments, safety benefits may come at the expense of congestion relief (i.e. speed and space optimization) and perhaps even efficiency.
Many of the issues and contradictions above may be engineering and machine-learning problems to solve, and AV enthusiasts will no doubt express confidence in autonomous technology’s ability to negotiate these challenges in a way that can optimize everything at once.
However, the important lesson from Walker’s rebuttal of O’Toole’s argument is the reminder that technology is not a panacea, and nor is it a solution in and of itself. A deductive analysis of a particular technology’s value for urban environments is a far less useful analytical framework than an inductive approach that begins with a problem and describes potential solutions, before then looking at how technology can help. In the case of AVs -- a technology which is not yet implementable -- this means thinking more clearly about the broader set of urban challenges and then describing a future where cities can successfully address as many of them as possible. Perhaps AVs are part of that future, if we think that their participation in public spaces can be a net positive, relative to all of the other challenges cities face. But we certainly should not start these conversations from the passive presumption that AVs are coming and the best we can do is prepare (or even celebrate, as O’Toole wants to do). Planners and officials -- local and national -- must decide how to adapt AVs to fit the cities we want, rather than retrofit our cities to adapt to AVs.
Edited by Justin Sherrill
Buried within an extremely interesting debate in September at the Cato Institute between Jarrett Walker (Oregon-based transit consultant) and Randal O’Toole (Oregon-based anti-transit writer) is a brief discussion of autonomous vehicles. Will AVs be an “extinction-level event” for transit, as O’Toole claims, by providing residents with a more convenient way to get around in a far less congested road environment? Or will their overall impact depend upon the precise nature of their deployment (which may still be years away) and the specific problem they’re trying to solve, as Walker argues?
For O’Toole, the problem is relatively simple: (1) people love driving; (2) driving is often too difficult or takes too long; and (3) we should therefore make driving easier. The personal automobile was a transformative and liberating event for humanity, in his view, and national mode-share trends -- despite billions of dollars in transit subsidies to support and expand systems with falling ridership -- demonstrate an overwhelming and consistent preference for driving. O’Toole argues that we should therefore optimize our transportation system for people’s expressed preference -- i.e. make driving easier and more convenient in a way that can also generate new transportation trips (which then creates new economic activity as well). He believes that AVs can achieve this outcome by reducing energy inefficiency, congestion, and roadway crashes.
Walker flips this logic on its head, and tries to first disentangle the myriad of transportation challenges cities experience as a way to think more clearly about the specific role for new technologies (rather than start from the normative view that cars should determine how we plan for the future). For example, efficiency, congestion, and safety are all distinct problems that may require different solutions.
Can AVs address all of them? Well, the problem of energy efficiency could be addressed through electric vehicles. But what happens if those electric vehicles become AVs that run more often and therefore require more regular battery charging? Sustainability and efficiency at that stage depend upon the sourcing of the energy from the power grid -- and we are still, unfortunately, far away from 100% renewable energy at that level.
The problem of congestion is fundamentally a problem of space -- and urban environments are in many ways defined by limited space relative to the number of people (i.e. density). Can AVs optimize the use of this limited space in a way that significantly reduces congestion? That may depend upon another factor -- induced demand. If we make driving easier, then more people are likely to choose that mode. So while AVs may optimize space between vehicles and allow for somewhat faster (or more consistent) speeds, induced demand could easily lead to dramatic increases in the number of AVs on the road. Optimization of space is still constrained by the overall geometry, and more cars in the same roadway could offset the benefits of overall optimization. In these cases, large vehicles with multiple people could still be a solution (i.e. transit), and perhaps automated buses would be a more useful form of AV technology than cars -- particularly since they would follow fixed routes, which helps with optimization.
How about safety? According to data from the National Highway Traffic Safety Administration, nearly half of traffic-related fatalities occur in urban environments, with one-fifth of all fatalities on urban arterials alone. Since 90 percent of all motor vehicle crashes are caused by some form of human error, there could be an opportunity for AVs to make a dramatic difference in roadway fatalities. However, safety also means slower speeds to prevent sudden braking or swerving in response to unexpected conditions. What sort of unexpected things happen in streets and public spaces? Well...pedestrians, cyclists, scooterists, and others on active transportation will continue to use our city streets, and human behavior is unpredictable. AVs will have to account for the tremendous amount of street-level activity that cities generate, which often means slower, more careful driving, particularly within dense portions of a city’s grid network with shorter distances between intersections and high levels of mixed-use development. In these environments, safety benefits may come at the expense of congestion relief (i.e. speed and space optimization) and perhaps even efficiency.
Many of the issues and contradictions above may be engineering and machine-learning problems to solve, and AV enthusiasts will no doubt express confidence in autonomous technology’s ability to negotiate these challenges in a way that can optimize everything at once.
However, the important lesson from Walker’s rebuttal of O’Toole’s argument is the reminder that technology is not a panacea, and nor is it a solution in and of itself. A deductive analysis of a particular technology’s value for urban environments is a far less useful analytical framework than an inductive approach that begins with a problem and describes potential solutions, before then looking at how technology can help. In the case of AVs -- a technology which is not yet implementable -- this means thinking more clearly about the broader set of urban challenges and then describing a future where cities can successfully address as many of them as possible. Perhaps AVs are part of that future, if we think that their participation in public spaces can be a net positive, relative to all of the other challenges cities face. But we certainly should not start these conversations from the passive presumption that AVs are coming and the best we can do is prepare (or even celebrate, as O’Toole wants to do). Planners and officials -- local and national -- must decide how to adapt AVs to fit the cities we want, rather than retrofit our cities to adapt to AVs.
Edited by Justin Sherrill
Comments
Post a Comment