Autonomous Vehicles

Transit Leap: What Autonomous Vehicles Can Do for Transit

Massive fleets of shared autonomous vehicles will be realized more quickly by starting now with constrained geographic applications rather than focusing on SAE (Society of Automotive Engineers) levels of increasing robotic capability and accommodating partially-robotic vehicles one-customer-at-a-time.

Growing numbers of people following autonomous vehicle (AV) development infer that the days of the unreliable human driver are numbered. They debate how and when, but the number of doubters declines.

Once pervasive, AVs could bring huge benefits to cities and society, as long as their sheer volume does not overwhelm urban road space. Unfortunately, high volumes are the likely default.

A subset of AV enthusiasts sees potential for massive fleets of shared vehicles – robo-taxis, robo-shuttles, and the like. They expect this to virtually extinguish demand for personally-owned vehicles.

Recently, a half-dozen computer simulations suggest that ubiquitous, shared robotic vehicles could provide more cost-effective, universal, and equitable mobility than the current reliance on privately-owned vehicles. Vehicle kilometers traveled (VKT) would drop, but the incessant human demand for motorized mobility expressed as personal kilometers traveled (PKT) would continue to grow. This emerging path of development is preferable, but much work over multiple decades is needed to realize this.

Sharing is not the default mode of transportation

Some observers assume that virtually everyone would choose or even prefer to share instead of owning, and even see a moderation in PKT growth. Current research from the CityMobil2 group is a notable and balanced exception. While they see a modest shift in portions of people sharing vs. owning, they do see a rise in PKT.

The considerable optimism expressed by most sharing-economy observers is based on early trends (hovering well below 1 percent in the total VKT equation) among young, single, urban digerati in developed countries along with a dose of the commonsense rationality that behavioral economists caution against.

Only a tiny fraction of automobility observers think about the historical 20-year doubling time of the worldwide automobile population. People such as Dan Sperling, Deborah Gordon (Two Billion Cars), and Bill Ford Jr. are among these.

Shared vehicle fleets could indeed play a significant role – far above 1 percent – in reducing the number of vehicles needed to provide the equivalent PKT for this quadrupling of VKT. Planners need to begin the hard work of figuring how to arrange those fleets so that the required world population of extant motor vehicles 40 years from now will not be four times the size of our current fleet.

The three key issues – robo-driving, shared fleets, and VKT quadrupling – deserve equal time. Planners need to look beyond the current fascination with Feature Creep in automotive robotics, the untested assumption that drivers will all become natural sharers, and the near silence regarding a future where PKT demand manifested as VKT could readily overwhelm robotic technology including any sharing.

Transit Leap: A planned future for autonomous vehicles

In contrast, planners can start an immediate public-private community focus on ways to initialize and nurture the growth of Transportation as a Service (TaaS). Pioneers at CityMobil2 and the associated companies supplying practical lite-transit vehicles to these trials are showing the way.

The Transit Leap paradigm of gradual deployment is illustrated in the figure. AV-based service networks are installed and managed for constrained public applications, representing advancements in public transit.

Starting with short, closed-loop applications, moving to on-demand branched-routing applications, then robo-vehicle go-anywhere networks (taxis, vans, mini-buses) in ever-widening urban geographies would provide for gradually expanding reach for travelers. Over a few decades, complete urban regions would be serviced. Eventually, fleet operations would saturate megaregions, and then entire nations over the next 30 to 40 years.

Intentional future of sustainable transportation

Intentionally seeking public benefit via strong community action toward deliberately developed, shared autonomous fleets is better for all three dimensions of sustainability. Transportation thought leaders must avoid assuming or hoping the default path of encouraging and regulating increasingly-automated vehicles one-customer-at-a-time will undergo a miraculous, entrepreneurial transition to a path of widespread sharing.

Partnerships for Transit Leap need to be started now.

What do you think about the concept of Transit Leap? Would you prefer autonomous shared fleets or household autonomous vehicles? Share your opinions in the comment section.


Please note that this article expresses the opinions of the author and does not reflect the views of Move Forward.

12 Comments

  • John Niles
    25. November 2015 at 14:49

    A poster Bern Grush and I prepared for illustrating the concepts in this essay for display at the MIT Disrupting Mobility Global Summit on November 12 in Cambridge, Massachusetts, and at the Florida Automated Vehicle Summit on December 2 in Jacksonville, Florida, is posted in PDF at http://endofdriving.org/2015/11/12/application-creep-autonomous-vehicles-and-transit/ . Very recently, Bern and I have decided that “Transit Leap” is a better description of what is needed for worldwide sustainable mobility than “Application Creep.” Both of these terms are meant to stand in contrast to the long-established evolutionary concept in the automobile industry of “Feature Creep.”

  • opbrid
    16. December 2015 at 9:10

    I’m not sure I’d look to the the majority of existing Public Transit Agencies as leading this leap, our experience with them in the electric bus industry is that they are more cautious and slow moving than even the general public when adopting new technologies or paradigms. Years of testing, followed by protracted rollouts within existing contracts isn’t exactly a “Transit Leap”. Better to look towards private companies like Uber to lead the way with robo taxis, otherwise you will be waiting a very long time.

  • Stephen_Rees
    16. December 2015 at 16:06

    What is depressing about this article is that ignores land use – which is what drives transportation demand. Our current North American land use pattern, based on auto mobility, is fundamentally unsustainable. The authors concentrate on personal transportation as though it is a subject of itself and not intimately involved with how and where we live. We need better urban places where car use is not essential to every activity and there is greater opportunity for face to face interaction. The greatest improvements seen in recent years have involved reducing the amount of space in cities for moving – and parking – cars and replacing that with places where people want to stop as opposed to get through as quickly as possible. Until “systems and transportation engineers” become cognisant of urbanity and its importance to human happiness, vacuous proposals like this will continue to to be bandied about. I have no desire at all to live in the sort of place where everything I want to do requires me to get into a self driving automobile. Have you ever considered the use of the word “sprawl” in this context?

  • Grush Niles
    16. December 2015 at 16:27

    Thank you for your comment. Our long-run policy target is to quadruple the personal miles traveled (average PMT) per vehicle on the planet, to blunt the trend of vehicle population expansion, now projected by mid-century to be four times the 2010 count. We estimate that our target requires 75% to 85% of all person trips in shared vehicles, whether publicly or commercially operated. If the entire shift were made by public transit authorities, that would mean roughly a 3000% increase in current transit ridership, likely impossible. However, if most of the shift were to occur instead via TNCs such as Uber, then we would be concerned that equitable treatment for all income levels is threatened. We don’t want the financially disadvantaged to lose their ability to be mobile. Hence the full program of Transit Leap will have to be shared. We also caution that if public agencies are the laggards—as you suggest, and we fear—then transit workers are under threat of unemployment–especially non-rail. For all these reasons we work with government agencies, TNCs, P3s, innovative start-ups, co-ops, associations, and unions with equal enthusiasm. They are all needed. We are agnostic re how the work of reducing the required extant fleet will be shared among players.

  • Grush Niles
    17. December 2015 at 5:58

    Transit Leap is an operational concept for keeping public transit affordable and thus sustainable where land use pattern and density yield a degree of sprawl that makes traditional public transit impractical. Automation will eventually make transit vehicles of all sizes operate at a lower cost and thus extend the applicability of public transit to off-peak periods and lower density geographies. Transit Leap for efficient mobility along with parallel, simultaneous innovations in the basic utilities of energy and water supply, waste disposal, communications, food supply, health care services, public education, household non-food retail supply, and emergency services could make a greater proportion of the current North American land use pattern sustainable. Transit Leap is going to be part of making land use pattern and density somewhat less critical in sustainable transportation system design, thus making higher urban density more of a citizen’s choice than a requirement, while still subject to non-transportation considerations.

    The alternative to Transit Leap “Feature Creep” would generate far more sprawl… See slides 50 and 51 from: http://endofdriving.org/wp-content/uploads/2015/12/TransitLeapMississaugaMoves20151109_Final-2up.pdf

  • Martin Herraiz
    11. January 2016 at 23:34

    Why shared vehicles should only be either publicly or commercially operated? Wouldn’t it be better to share travels in our private cars? (considering that most of them run with several empty seats in almost every travel).
    Isn’t it better to harness what is already available (just using more efficiently our millions of private owned cars), than waiting for new technologies pending to be developed in the years to come?
    Please, tell me in what sense would a robo-car (full of sensors, radars, servo systems and the like top technology) be cheapest than a regular one?
    Isn’t it better to develop a simple mobile app to coordinate passengers and drivers for ride sharing today?

    To think that VKT=PKT is a mistake, because it would mean that every car carries just one person (the driver). The fact is that most cars have 5 seats and can carry 4 passengers in addition to the driver.

    Self driving cars could make things even worst (VKT>PKT) every time that they run with nobody inside.

    The intelligent way is to make VKT<

  • Martin Herraiz
    11. January 2016 at 23:34

    Why shared vehicles should only be either publicly or commercially operated? Wouldn’t it be better to share travels in our private cars? (considering that most of them run with several empty seats in almost every travel). Isn’t it better to harness what is already available (just using more efficiently our millions of private owned cars), than waiting for new technologies pending to be developed in the years to come?
    Please, tell me in what sense would a robo-car (full of sensors, radars, servo systems and the like top technology) be cheapest than a regular one? Isn’t it better to develop a simple mobile app to coordinate passengers and drivers for ride sharing today?
    To think that VKT=PKT is a mistake, because it would mean that every car carries just one person (the driver). The fact is that most cars have 5 seats and can carry 4 passengers in addition to the driver.
    Self driving cars could make things even worst (VKT>PKT) every time that they run with nobody inside.
    The intelligent way is to make VKT<

  • Grush Niles
    21. January 2016 at 6:35

    Martin, we are espousing lower levels of ownership. As low as possible. Ideally only service vehicles (plumbers, electricians) would be owned by individuals. The fewer owned vehicles, the less of the sharing you espouse would be possible. Also, the fleet we have today will be gone and replaced by the 2035 or 2045 fleet(s). The whole world of cars you see out your window will change for better or worse. It will not be the same. Indeed, by 2050, the world is slated to have 4B of them instead of the 1B we had in 2010. We think it is not fruitful to map AVs onto today’s use/cost/energy/ownership models. If you do, you should be truly frightened.

    Otherwise, autonomous vehicles would be in commercial or municipal fleets and would be shared: both serial (carshare-likely favorite) and parallel (rideshare-your favorite). Studies so far indicate that there is much more potential for serial sharing than for parallel. Why: for parallel sharing you need five points of alignment: [1] proximate origin, [2] proximate destination, [3] proximate departure time at origin, [4] willingness to share, and [5] an available vehicle for proximate departure time. For serial sharing, you only need one thing: an available vehicle. We absolutely agree ridesharing is superior to carsharing, CO2-wise, but uptake is and will remain limited by comparison.

    In 2035, robocars will be cheaper (in 2016 dollars) that 2016 cars are today: Moore’s Law, 3D Printing, super materials, energy internet, lower insurance, no parking, more sharing, and others. Read Rifkin’s Zero Marginal Cost.

    And you’re right the AV could make things worse. This the core reason for GrushNiles’ work. Glad you see that. Many, but not all see that potential. We suspect that if left to its own devices, the AV will indeed make it worse. (see Feature Creep, in the PPT at the link below, 2nd or 3rd last slide)

    Also see this: http://endofdriving.org/2015/11/12/transit-leap-autonomous-vehicles-and-transit/

  • Martin Herraiz
    21. January 2016 at 23:41

    We, the people, love cars due to the sense of freedom they give us. They are beautiful machines and, as BMW advertising says: “I love driving”. In developing countries, everyone is expecting to own one. What we don’t love is to have to use them everyday to go to work, and to have to bear their expenses alone. In this sense, parallel/ridesharing can help a lot.
    Of course, different and complementary approaches to traffic problem solving can overlap, addressing different aspects of the matter. I agree that serial/carsharing might contribute to solve low use of cars, that stay parked for long time; but definitely it will not help much to reduce the number of cars required to carry people. Do you know why? Because of rush hours exist in people behaviour.
    I’ll explain to you with an example: In my home town, Madrid, you are allowed (and encouraged) to bring your bike in the subway all day except for certain two hour long time slots: around 8:30, 15:00 and 20:00. The rest of the time there is plenty room in the trains, but at those peak hours the trains are crowded. Let’s translate this phenomena to a serial/carsharing solution: There will not be enough of them at peak hours, but most of them will be parked (or even worst, running with nobody onboard) the rest of the time.
    I appreciate your clever definition of the five point of alignment required for parallel/ridesharing. I have been thinking for four years about how to solve this problem and on how to increase chances for passengers to find a ridesharing car going their way (and for the car to find passengers) and I finally arrived to a procedure that avoids most of them (but willingness to share): The key is not the trivial origin/destination approach, but the flexible route approach. I will be glad to explain better. For the time being let’s say that the procedure is contained in this patent: https://patentscope.wipo.int/search/en/detail.jsf?docId=WO2015110677&redirectedID=true

  • Martin Herraiz
    21. January 2016 at 23:49

    We, the people, love cars due to the sense of freedom they give us. They are beautiful machines and, as BMW advertising says: “I love driving”. In developing countries, everyone is expecting to own one. What we don’t love is to have to use them everyday to go to work, and to have to bear their expenses alone. In this sense, parallel/ridesharing can help a lot.
    Of course, different and complementary approaches to traffic problem solving can overlap, addressing different aspects of the matter. I agree that serial/carsharing might contribute to solve low use of cars, that stay parked for long time; but definitely it will not help much to reduce the number of cars required to carry people. Do you know why? Because of rush hours exist in people behaviour.
    I’ll explain to you with an example: In my home town, Madrid, you are allowed (and encouraged) to bring your bike in the subway all day except for certain two hour long time slots: around 8:30, 15:00 and 20:00. The rest of the time there is plenty room in the trains, but at those peak hours the trains are crowded. Let’s translate this phenomena to a serial/carsharing solution: There will not be enough of them at peak hours, but most of them will be parked (or even worst, running with nobody onboard) the rest of the time.
    I appreciate your clever definition of the five point of alignment required for parallel/ridesharing. I have been thinking for four years about how to solve this problem and on how to increase chances for passengers to find a ridesharing car going their way (and for the car to find passengers) and I finally arrived to a procedure that avoids most of them (but willingness to share): The key is not the trivial origin/destination approach, but the flexible route approach. I will be glad to explain better. For the time being let’s say that the procedure is contained in this patent: https://patentscope.wipo.int/search/en/detail.jsf?docId=WO2015110677&redirectedID=true

  • Paul Minett
    11. February 2016 at 3:11

    I think both will happen.

    There will be people who own autonomous cars that, by their nature (the cars) will easily be absorbed into a fleet when they are not in use by the owner, and by their nature (the owners) will be pleased to have a revenue stream that pays for the car. This ‘distributed capital’ model will make it easier for fleets to get established. Their precursors can be seen in the likes of GetAround and RelayRides.

    Then there will be fleets. Big money will buy fleets and operate them, in the same way that Car2Go operates at the moment. The difference will be that the Car-share-car will come to you, rather than you going to it.

    Uber and their ilk might be well placed through existing customer relationships to lead the transition from driven to driver-less cars, with the customer able to specify at booking if they have a preference for a driver or none. Of course, at the moment Uber is not a big owner of cars, and if they maintain their current model it might be that they enable the fleet of privately owned autonomous cars funded with household capital.

    In the discussion about this post, Grush posits that ‘parallel’ ridesharing will not be a big part of the solution, which commenter Martin posits will therefore limit the overall reduction in available fleet size due to peak demand. The authors should pay attention to the apparent success that Uber Pool and Lyft Line are having – it seems that there is a level of acceptance of being a sharing passenger that exceeds Grush’s expectations. Note that when you share in an Uber Pool, your five points of alignment become moot. Gone is the need for a common origin, common destination, common departure time, and it seems that there is a higher willingness to share than is usually expected.

  • Grush Niles
    14. February 2016 at 23:18

    Paul: John and I are aware of the *publicized success* of Lyft Line and Uberpool. Indeed, this is a promising development. Our thinking that ride sharing has limitations comes from a few sources.

    First is the disclosure in the Kockelman-Fagnant *simulation* paper *Dynamic Ride-Sharing and Optimal Fleet Sizing for a System of Shared Autonomous Vehicles* (TRB, 2015, page 8) that “just 6,152 ride-sharing matches out of 56,324 trips” (about 11%) in a population with which *every trip* starting within a reasonable time-distance proximity was ride shared. This study simulated *time/space opportunity* and neither stated nor revealed preference. We assert that a stated preference would be lower and a revealed preference lower still. Our sense is that of every traveler whose trips matched the simulation criteria perhaps half or less (we are guessing less) would accept the offer for various behavioural economic reasons.

    Second, in some cultures and among some social groups sharing is less accepted than among others. Sharing with strangers, even less. We grant that our five points of alignment are relaxed by the Lyft-Uber programs but would argue they are not mooted. Instead of sharing a proximate origin, the second-trip origin need only be proximate somewhere on the trip path. Same with a proximate destination and time alignment.

    In the end we are *guessing* (not even asserting) that ride-sharing will play only a minor role compared to carsharing.

    Aside from Bern:
    I grant that to be wrong in this case would likely be a good thing. So I hope you are right and we are wrong. I don’t even know how to wager this Paul, since I also cannot guess the value of bitcoins by then. (Hint: I think we will see the end of cash before we see the end of people’s fear of other people.) Let’s bookmark this and check back in 2035.

Related

There’s plenty of buzz around autonomous vehicles being the future of mobility – every commuter’s saving grace. It’s glamorous to imagine drivers relaxing behind the wheel while their cars navigate rush hour traffic without any human assistance. What’s less glamorous is the gridlock itself – a larger issue that won’t... View Article

For most, the thought of riding in a driverless car is hard to imagine. Although five years ago it may have seemed improbable to press a button on your smartphone to instantly summon a private car driven by a stranger – and actually trusting that driver to get you to... View Article