RE: [TMIP] Re: [TMIP] VDFs for autonomous vehicles

1 post / 0 new
RWFarley
RE: [TMIP] Re: [TMIP] VDFs for autonomous vehicles

Scott,

I think we've been seeing this play out. Areas with higher recurring congestion have been experimenting with BPR functions using a higher Beta value. This could be interpreted as the effect of more aggressive drivers "pushing through" while under capacity. The downside is that once you hit capacity, the breakdown is worse.

I have been of the opinion that 100% highly capable AVLs would change the Delay Curve into a Heavyside function. Great ... up to a point.

I think there are too many unknowns in terms of how this will (be allowed to) play out. As AVLs are introduced, what will the populace do if a 6,000 car "train" cuts off access to sections of the city every morning? Will there be "mandatory" weaving allowances? Lots of questions, and this is only assuming a User Optimal market as we have now. My feeling is that as one or another effect becomes troublesome, the city will react (possibly well, possibly poorly) and "correct" the troublesome behavior. This would have the effect of keeping things relatively consistent. In a heterogeneous market, AVLs would need to "seem like normal cars" to drive safely and efficiently. If they become overly aggressive, the majority of "normal car drivers" will demand they "behave."

After 100% AVL (at least 40-80 years out) will we decide to impose a System Optimal market? Can I sell a faster trip for you and a slower trip for me by swapping paths?

And in any case, the real question is, how will people "game the system" for their personal advantage?

Thanks

From: scott.thomson=ky.gov@mg.tmip.org [mailto:scott.thomson=ky.gov@mg.tmip.org] On Behalf Of .Scott
Sent: Friday, April 27, 2018 08:21
To: TMIP <tmip@mg.tmip.org> Subject: [TMIP] Re: [TMIP] VDFs for autonomous vehicles

Another possible expectation is that autonomous vehicles will be far more aware

of operating efficiencies which would further erode travel times.

It's the less aggressive behavioral changes in traffic flow characteristics, especially in the earliest period of mild market penetration that may actually lead to higher incident rates.

Scott Thomson, P.E.

Model Team Lead

Division of Planning

Kentucky Transportation Cabinet

502-782-5086

________________________________
From: amirshahpar=gmail.com@mg.tmip.org<mailto:gmail.com@mg.tmip.org> <amirshahpar=gmail.com@mg.tmip.org<mailto:gmail.com@mg.tmip.org>> on behalf of qykamalyg <amirshahpar@gmail.com<mailto:amirshahpar@gmail.com>> Sent: Friday, April 27, 2018 7:40 AM
To: TMIP
Subject: Re: [TMIP] VDFs for autonomous vehicles

Jill,

You bring up a very important point. We worked on this hypothesis that AVs
in a mixed environment maintain a safe distance from the vehicle following,
and that distance might well become an acceptable gap for human drivers in
heavily congested corridors where they become more aggressive, risk takers.

We created a test scenario including a typical weaving segment of a freeway.
VISSIM software was used and travel times for the main movements where
measured under mixed environment and moderate congestion (V/C~~0.95). Our
simulation showed that low (<40%) AV market penetration can lead to increased travel time by 8 to 20 percent. The average speed reduction for AVs was shown to be between 30 to 80 percent in the mixed environment compared to 100% AV system. Thanks, Amir *Amir Shahpar, P.E. | *Modeling Manager / Senior Planner | Transportation Planning & Investment | Virginia Department of Transportation Office: 703-259-1731 | 4975 Alliance Drive, Fairfax, VA 22030 On Thu, Apr 26, 2018, 1:40 PM jill hough wrote: > Suzanne, et. al.,
> In addition to the many points being made, has anyone explored
> possibilities that in a 100% AV system (or even within mixed fleet
> scenarios), the speeds of AVs will be considerably slower? With emphases on
> 'reliability' and safety, combined with the vigorous reporting of accidents
> due to av testing, combined with the likely decreases in VOT that the
> user/occupant may have (in that they can consume the time in the vehicle to
> be productive in some way), would there be a tendency for speeds on roads
> to erode in response? It will be interesting to see of the micro-simulator
> work in Chicago will be reflective of this.
> Jill
> > On Tue, Apr 24, 2018 at 3:40 AM, Mahmoud Javanmardi
> wrote:
> > > Suzanne,
> > > > Our team at Argonne National Lab has been working with Texas A&M
> > University (Dr. Talebpour and his team) to generate the VDFs at different
> > market penetration rates using their micro-simulator. The micro-simulator
> > uses behavioral car following and lane changing models to control regular
> > vehicles and ACC/CACC logic or deep learning (trained with actual AV
> data)
> > are used to control AVs. For the Chicago region, we have identified 60
> > clusters to generate the VDFs for, and we will publish the challenges and
> > results soon.
> > > > Mahmoud Javanmardi
> > Principal Computational Transportation Engineer
> > Systems Modeling and Controls
> > Argonne National Laboratory
> > --
> > Full post: https://tmip.org/content/vdfs-autonomous-vehicles

VDFs for autonomous vehicles | TMIP FMIP tmip.org
Hello modeling friends, Has anyone tested or imagined any reasonable assumptions for VDFs in all-autonomous vehicle scenario? Or, if not, does someone have some good

> > Manage my subscriptions: https://tmip.org/mailinglist > > Stop emails for this post:
> https://tmip.org/mailinglist/unsubscribe/12676 > > > --
> Full post: https://tmip.org/content/vdfs-autonomous-vehicles > Manage my subscriptions: https://tmip.org/mailinglist > Stop emails for this post: https://tmip.org/mailinglist/unsubscribe/12676 >

--
Full post: https://tmip.org/content/vdfs-autonomous-vehicles Manage my subscriptions: https://tmip.org/mailinglist Stop emails for this post: https://tmip.org/mailinglist/unsubscribe/12676 --
Full post: https://tmip.org/content/re-tmip-vdfs-autonomous-vehicles
Manage my subscriptions: https://tmip.org/mailinglist
Stop emails for this post: https://tmip.org/mailinglist/unsubscribe/12691