CAV Workshop Discussion

go back to Event Site

 

Questions Raised During CAV Workshop Discussion

 

To what extent is our focus on passenger versus freight modeling?
Can the slides for all of the presentations be provided to everyone?
At least as important as the technical challenge to build AVs is the question how quickly users will accept this new technology.
Accessibility and land use choices interact. Will persistence in current land use shape CAV or will CAV shape land use. What land use patterns are we assuming?
Looking at vehicle sharing, are we focusing enough on issues like the value of diversity in vehicle choice and the symbolic value of vehicle ownership?
Responding to Jeremy Raw: I agree with your definition of models & scenarios. I believe that our task is to move more aspects from scenarios to models by developing better travel behavior theory.
Tom Rossi, if we need to model many scenarios, what are the implications for model runtimes? Overnight run? Two hours runtime?
Even if trqvel time disutility is reduced does it mean VMT goes up if demand is already satisfied  where is the inducement
Rolf, the time to run the original model depends on the model, of course.  In TMIP-EMAT, the Buffalo model took something like 4-6 hours to run, and I think we did 100 runs to create the metamodels.  But the metamodels themselves take only a few seconds, and so we did thousands of them.
A larger model could take much longer; even with multiple servers it could take a while.  I think that is why all of the agencies that have done scenario based analysis have done only a few scenarios.
Do we have data from areas with high TNC mode share (e.g. San Francisco) on how much pooling is happening?
Is everyone aware of ubermovement? a data website.
I’m wondering whether we need to distinguish between the research that we all probably agree is needed and how planners who need to plan now can best use what exists or what can be created quickly.
Have academics reviewed the work of Alex Bayen at UC Berkeley on Flow?  Any contrarian findings?
Can AVs be adequately represented in HCM analysis of uninterrupted flow facilities?  Is the Highway Capacity Manual ready for AV scenario analysis?
What about modeling for roundabouts?
In modeling capacity, what changes are needed to account for potential variation in vehicle sizes?
What is the operational effect on arterials of converting parking to AV pick up/drop off?  Do the extra lane transition events reduce throughput?
What might be the new factors leading to travel (and deterring it) if travel time ceases to be the major determinant?
Should the group also consider the effects of Automated Transit?
How will AV effects influence tolling/congestion pricing effectiveness?
Does the number of TNC or AV fleet operators influence the system performance especially zero passenger VMT?  For example, is a monopoly provider preferred to minimize wasted VMT?
A commercial enterprise deciding on who are likely buyers, or voters, will use a dozen variables to categorize the market and  propensity to “buy.”  What travel model uses such extensive segmentation to inform traveler choices.
To add to Ron's question, TNCs using AVs may operate differently than today's situation where drivers may work for multiple TNCs.  If the TNCs own the AVs, there might be more vehicles, and therefore more deadhead VMT.
I need to put AVs into my mode choice model because I care about the competition with other modes.
Norm Marshall's comment reminds me that network services tend to be natural monopolies with declining marginal costs across all firm sizes. But, does this need to be modeled? Are the competitive issues (efficiency vs. market power) of sufficient interest to public policy?
With human drivers, a lot of the "empty"vehicle miles are to from the drivers' homes - an extra "commute" trip - that would be avoided with AVs.
How does a service such as Hop Skip Drive affect a household’s auto ownership decision?
What are the important factors influencing the auto ownership decision when the annualized cost of TNC/AV service equals private ownership?
Can we predict the accessibility effects of AVs if we can’t estimate the speed of AV travel given current uncertainties about network performance?
How does the cost of providing commercial AV travel service change when consumers demand different types of vehicles? The idea is that offering a diversity of vehicle types for trips decreases the utilization rate.
How can we separate vehicle ownership from use as various mobility plans (subscription plan) are finding their way into the market?
In the future, apartments might offer access to AVs that belong to the apartment building.
California will add 10 million people by 2050. At our current urban densities it will take about 2,000 square miles to house, employ, and serve this population. This is a land mass larger than the current cities of Los Angeles, San Francisco, Chicago, New York, Atlanta, and Dallas. What then?
completely agree re: residential developments offering AVs for common use.  In the US, retirement communities would be a prime target (many today offer shuttle services; AVs would offer high level of service, once trust barrier is overcome).
Could reduction in parking requirements reduce higher-density development costs enough to at least partially offset sprawl impacts?
...separate vehicle ownership from use"Ownership is one factor that would affect use-- but the latter will also affect ownership. From modeling perspective this is a classic joint model structure for car ownership (eg discrete continuous models by Train and others; to be adapted).
wouldn't the fact that the vehicles are shared by multiple users increase utilization rate (for the vehicle), and for the users as well (since there would be a "best" vehicle for each type of use)?
re: accessibility and AV travel speeds.  This is no different than with existing models; demand patterns and network performance need to be determined in mutually consistent manner.
re: Empty VMT and number of TNC's.  Empty VMT decreases the more AVs an entity controls, and  the larger the addressable market, because of more efficient routing and vehicle-customer assignment. See: https://www.sciencedirect.com/science/article/pii/S0968090X18306028
Has anyone looked at how TNCs impact individuals residential location choices? This can be a good starting point.
On the land use and other long term choices, what’s the mechanism of land use and AVs / people’s travel behavior interact? Land use shapes our travel behavior or vice versa? Or it is constantly interacting relationship? It’s critical to figure this out...
Our study areas need to get larger to cover the area affected by AVs.