Autonomous Vehicles, New Mobility & the Built Environment

A Transition Step(s) to an Autonomous Transport Future

A relatively low-density city designed around the car. Public transportation service that meets its published schedule only 71% of the time, leading to customer satisfaction of only 65%, while the fare recovery is only 15% (passengers only pay 15% of the actual cost with government subsidies making up the difference). Although the above sounds like a description befitting most any U.S. post-World War II city, this is how the creator of a software simulation program and associated white paper describe Canberra, Australia’s current state of public transportation.

A 21st Century Approach to Transportation Design #

Kent Fitch of Project Computing, an Australian software company, states the problem if planners and politicians continue to design transportation networks with a 19th or 20th century approach:

“However, no incarnation of the current bus network nor even an extensive light-rail system can be imagined which meets these goals [decreased traffic, pollution, better health and safety, affordable and efficient public transportation], short of “starting again”: abandoning the 100 sprawling, far-flung suburbs of the “bush capital” for a high-density new city. Hence, both bus and light-rail are distractions from meeting the city’s transport goals, but are being pursued only because no better alternatives are perceived.”

Utilization rate for a low (120k journey per day) take rate.
Utilization rate for a low (120k journeys per day) take rate.

Fitch has created an impressive software simulation that allows anyone to set various parameters of what would be an autonomous vehicle public transit system. Of course it has the limitation of being a model, but the results shout out to planners and city leaders everywhere, that self-driving vehicles need to be a big part of the discussion of both the transport (public and private) and built-environment future.

As shown in his model, the benefits of autonomy increase, as more people choose shared autonomous transport over private vehicles. For instance, his worst-case scenario, where the number of journeys equals that of the current bus system, yields a 90% subsidy reduction, (A$120M to A$12M), elimination of the 92 grams of CO2-e per passenger km attributed to buses, while increasing peak-time congestion (which could be mitigated by increasing wait-time). As the number of journeys increases, costs continue to decrease, while congestion also decreases because of the shared nature of the autonomous vehicles.

‘Traffic congestion is dramatically decreased, particularly during peak periods on major roads. For example, the average occupancy of cars arriving is Civic and Parkes is 2 passengers, compared to an estimate of 1.13 for current journeys to work.”

His work reinforces studies from others (e.g. Fagnant) regarding improved transport efficiency and pollution reduction (elimination of tailpipe emissions) from sharing autonomous vehicles, as well as other societal benefits (Godsmark suggests $2,700 annual net benefit per household (PDF).

The challenge is the transition from today’s scheduled transport world, with its high-fixed costs, to one that is on-demand and where resources can scale much more closely match demand. How does one get the benefits without prematurely abandoning capital or increasing congestion along the way?

A First Step – Bus Feeder with Autonomous Last Mile Transport #

Image of an AuRo Robotics self-driving vehicle.
Image courtesy of AuRo Robotics

A scenario that could potentially address the transition and that would be a good candidate to add to Fitch’s model is one that combines the existing bus system with lower-cost, last mile autonomous transport. That is, existing, human-driven buses would be used as feeders on major thoroughfares. Bus stops between thoroughfares would be eliminated and replaced with transit plazas at the intersections of the major thoroughfares. Autonomous vehicles would provide last-mile transport.

Being unfamiliar with Canberra, Australia, this author examined a small portion of Silicon Valley to understand what these assumptions might mean:

  • Buses would not be eliminated, but routes would be streamlined, such that they would be on major thoroughfares generally traveling East-West or North-South. Bus stops between intersections (now, as often as six per mile) would be eliminated. These two things alone would reduce transit time for riders (fewer stops) and reduce the congestion caused by buses navigating their way into and out of traffic from bus stops. Of course, this approach doesn’t preclude the use of bus-only lanes, but even those type of lanes could be made more efficient by running in half-duplex mode, such that only one lane would be required for both directions (assuming 45 MPH and a mile between transit stations, then it would take less than 2 minutes to travel between plaza stations, which would easily allow up to a 5 minute frequency between buses.
  • Win6The ideal transit plaza would be a mixed-use, multi-level development that covers major intersections, such that it doesn’t interrupt the existing traffic flow, but provides a way for the autonomous vehicles to feed passengers to the aforementioned buses (e.g. the buses and autonomous shared ride vehicles would converge on a level above the traffic).

As much as these structures would be about improving traffic flows, they would also be about creating new land in a place where land is extremely valuable. An estimate suggests that 30 acres could be gained in one 3.2 square mile alone. With Silicon Valley land typically ranging from $1 to $4M per acre, this could mean unlocking $30 to $250M of value. Not all of these plaza sites would be able to take advantage of mixed-use and might have limited use (e.g. mini-parks, plazas), while others could support multi-story developments and include things such as housing, businesses and shopping.

Image Courtesy of Win6 Village
Image Courtesy of Win6 Village

Any housing and jobs created at these sites would be at ideal locations, as they would be at transit hubs. Additionally, the government entities that own the underlying land could consider creative ways to create affordable housing above these publicly owned air spaces.

  • Autonomous last mile transport is assumed using vehicles such as the one from AuRo Robotics that is being tested with real-world traffic at Santa Clara University. These type of vehicles would most likely be even lower cost than the $40,000 per vehicles cost that Fitch assumed in his model. Given that these are lower speed vehicles (25 MPH maximum), safety is enhanced and it would more easily pass regulatory hurdles than a car designed for highway driving. AuRo Robotics suggests 2016 for commercial introduction for their vehicles, so autonomous vehicles, in some form, will probably be available before society at large is ready.
Image courtesy of AuRo Robotics
Image courtesy of AuRo Robotics

These sort of vehicles could impact the built environment, freeing up more land for use by people and not their cars. That is, with larger developments, roads for traditional vehicles could be eliminated or reduced. Parking for private cars could be at the aforementioned plazas or at the edge of a development (away from housing). These smaller autonomous vehicles would ferry people and their goods within the development.

Government Transit Agencies Should Build on Fitch’s Work #

One of the really cool things about Fitch’s model is that it is licensed under a  Creative Commons Attribution-NonCommercial 4.0 International License. That means that anyone (with the right kind of software knowledge, of course) can adapt Fitch’s model to see how variations, such as the one given above, impact the economics and quality of public transport. This is something that transit agencies should be doing as a matter of course to ensure that they are using the public’s money wisely.

Of course, it wouldn’t be surprising to see  private companies jump in on this opportunity. For instance, Alphabet (i.e., Google) already runs a bus fleet for their employees and they have their well publicized autonomous vehicles. As written previously, mobility as a service could be a significant business rivaling its traditional search business.

Regardless of whether it is government, private enterprise or combinations of public-private partnerships, autonomy is set to disrupt mobility. Kent Fitch’s model is a good starting point in understanding the potential impact of this disruption.


Author Ken Pyle, Managing Editor

By Ken Pyle, Managing Editor

Ken Pyle is co-founder of Viodi, LLC and Managing Editor of the Viodi View, a publication focused on independent telcos’ efforts to offer video to their customers. He has edited and produced numerous multimedia projects for NTCA, US Telecom and Viodi. Pyle is the producer of Viodi’s Local Content Workshop, the Video Production Crash Course at NAB, as well as ViodiTV. He has been intimately involved in Viodi’s consulting projects and has created processes for clients to use for their PPV and VOD operations, as well authored reports on the independent telco market.

Linked In Profile

25 replies on “A Transition Step(s) to an Autonomous Transport Future”

Here was Toyota’s 2013 vision of how multimodal transit might work. In many ways, what they are talking about is Waze. It does a good job of illustrating how multi-modal transit, mixed with intelligence throughout the network, could create a much more efficient transportation system than today.

And this study of Bogota’s Bus Rapid Transit system. As can be seen, successful BRT requires more than optimized routes, but it also needs stations that are designed like metro stations (e.g. flat ramps to get onto the bus, doors that open when a bus arrives, sheltered from the weather), special buses (e.g. that have doors on the left-hand side) and pavement that is strong enough to support high-capacity heavy buses that carry lots of passengers on the trunk lines.

And it looks like the UK might be investigating an approach that is similar to what is suggested in the above post

and the group that is behind this and their explainer video. The only thing it is missing is a cap

Here is the latest on a micro-transit solution in Johnson County, KS (a suburb of Kansas City), which emulates a MaaS, albeit with a driver.

They are spending about $500k over a year and are on track to provide about 35,000 on-demand, pooled rides. It’s an app-based solution (app provided by Ford subsidiary, Transloc) that costs the rider $1.50 per ride. The math says about $14+ is the actual cost/ride for about 7% fare recovery or about the same as Santa Clara County’s VTA. Presumably, this provides much better service than a regular bus line. 

It would be interesting to understand the average utilization of their 12-person vans, as well as how much of the $500k is actually fixed upfront costs versus variable costs.  

and the description of the service when it launched (note they were looking for between 100 to 150 riders per day, which they are achieving with 110 riders per day).  

Closer to home is Via’s $1.7M, 18 month pilot in Cupertino with discounted rides of $2.50 for youth, low-income, and seniors , while it is $5 for everyone else. Or, one may get a weekly pass for $17 (up to 4 rides per day) or a monthly pass for $60. With the exception of the all important Caltrain station in Sunnyvale, it only services the city of Cupertino, however, and has limited hours which are 6 AM – 8 PM Mon-Friday, 9 AM – 5 PM on Saturdays (no service on Sunday). It launches on 10/29 and doesn’t require a smartphone.

It will be interesting to see what sort of demand there is for this pre-autonomous vehicle ridesharing service.

And suburban Santa Clarita, which is only 18 miles or so from downtown LA, but feels like a 100 miles, is trying on-demand vans for first/last mile during peak hours ($1 reduced/$2 regular). Of course, as Dr. Kornhauser says, the only way this can scale is to remove the driver, like Waymo is doing in Chandler.

And Via makes a good case for micro-transit, ridesharing with data from around the world regarding their deployments. They show a 24 to 47% reduction in single occupancy trips with the deployment of their on-demand micro-transit solution. They don’t talk about it, but it will be interesting to see what they implement (e.g. dividers) to make people feel safe in the post-shutdown era..

Via further makes the argument that on-demand shuttles reduce costs, while providing better rider experience than fixed routes. In the pilot they are running with Sault Ste. Marie transit services they are using standard 40-foot buses for on-demand, instead of fixed routes on Sunday evenings., when demand is normally low.

This city of 73,000 is saving money while reducing service hours by 11% and decreasing wait time to 12-minutes from the typical one hour.

Arlington, Texas, a city of 400k, is reportedly planning on spending $9M per year to serve its entire 99 square miles with 70 shuttles operated by Via. Via’s on-demand shuttles are an alternative to traditional buses. Unlike buses, the Via shuttles pick up and drop off riders within 2 blocks of the rider’s location, making their solution much more accessible than traditional bus service with bus stops which tend to serve only major roads. The price to the rider is about $3 to $5 per ride.
With approximately 4,000+ square miles, Arlington is about 68% the density of the City of San Jose and is the largest city in the U.S. without bus service. Three times, citizens voted down tax proposals to fund a traditional bus system. In its initial trial, which covered 65% of Arlington, Via provided 112k rides and the cost from Via to the city was $1.7M.* As a simple metric, this translates into a subsidy of approximately $15.04/ride.
Compare this to Santa Clara County’s VTA, which serves a population of 1.9M, covers 363 square miles of incorporated cities. and had a fleet of 469 buses in 2019 (page 40).* VTA’s pre-pandemic project was approximately 26.3M bus rides (page 46). Fares account for about 7% of the budget (page 47). VTA’s subsidy per ride is approximately $16.60 to $20.60 per ride.*
Granted, at $3 to $5 per ride, the Arlington/Via cost to the rider is a bit more than $1-$2.50 per ride for VTA. Still, even if Arlington increased its per-ride subsidy to match the VTA fare (say $2.50 per ride), the subsidy would still be less than VTA, while providing a higher quality experience (on-demand, closer to origination/destination).

Long-term, subsidies might not be required if the operating costs drop to 1/5 of today’s cost, as predicted by Local Motors’ Jay Rogers.

Transit agencies boards would be well-served to closely look at the Arlington approach, as well as how that fits with the emerging driverless evolution.

*Although the county is 1,312 square miles, the incorporated portion is 363 square miles. The 14 smaller municipalities are 181 square miles, almost matching the largest, San Jose, in area, San Jose with its 182 square miles.
**This is based on $441M revenue (page 56) and 26.5M irides. It doesn’t include depreciation on capital costs, which it should as the biennial 2020/2021 budget appears to be $220M (110M per year) for capital improvements, repairs, and maintenance. This could add up to $4+ per ride, as it is a real cost. Granted, some of these might be for long-term capital, but there is no reason to think that capital expenditures will be lower in 2022/2023, unless their funding sources are cut.

It’s important to note, that Arlington doesn’t have capital costs with its approach. Their expense is variable, although there could be administrative costs not accounted for in the Gov Tech article.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.

%d bloggers like this: