Re: [sumo-user] fixing bugs in a jammed simulation

[Mirko Barthauer <m.barthauer@xxxxxxxxxxx>]

Hi Sohail,

 

it is difficult in any case to debug problems in larger networks, so I can give you only some hints where to look. Do you have any other config parts related to rerouting? I miss the --device.rerouting.period argument. As stated in the docs, otherwise "a vehicle gets a routing device only rerouting before insertion is active by default". So it doesn't take into account later changes in edge occupancy.

 

Best regards

Mirko

 

 

 

 

-----Original-Nachricht-----

Betreff: Re: [sumo-user] fixing bugs in a jammed simulation

Datum: 2023-10-04T06:52:07+0200

Von: "Soheil Keshavarz" <soheil99@xxxxxx>

An: "Mirko Barthauer" <m.barthauer@xxxxxxxxxxx>

 

 

 

Thanks again for your response. My next question is about some unreasonable routings. 

First of all, here is a screen of some part of my latest sumocfg file:

--- My vehicle properties:

  <vType id="passenger" vClass="passenger" accel="3.2" color="yellow" decel="3.5" length="4" maxSpeed="200" lcStrategic="5" lcSpeedGain="10" lcCooperative="1" lcSpeedGainLookahead="5" minGap="1.5" cc1="0.9" cc2="4" cc3="-8" cc4="-0.1" cc5="0.1" cc6="11.44" cc7="0.25" cc8="3.5" cc9="1.5"/>

 

Here I will share some screenshots about some unwanted behaviors of my agents. I'm looking for some suggestions to avoid these behaviors. I thought making weights.priority-factor from 0.5 into 8 should do the trick but the problem is still the same. 

1- In the picture below, some vehicles tend to choose a minor link and if they just keep driving in the left lane, which is almost empty, they will go through this link much faster. How can I avoid such undesirable routings?

 

2- Realistically, people traveling from O to D are expected to remain on their link and go through the green arrow. However, a big portion of them decide to reroute and use the links with lower priority in red arrow directions. It's worth mentioning that these vehicles will still end up in congestion so I'm not sure if they are saving time.

 

3- More interestingly, those vehicles will choose roads with lower priorities (red arrows, priority=5,6) and alleys (orange links, priority=2) rather than main streets(priority=7, green arrows). 

4- Even if some vehicles choose to take the left lane of the main street, they will turn into a jammed alley. Regarding both link priority and travel time, going into that alley is certainly more costly. 

 

Once again, I would appreciate the time you take to read and answer this email.

Thank you

 

Soheil

On Mon, Oct 2, 2023 at 1:31 AM Mirko Barthauer <m.barthauer@xxxxxxxxxxx> wrote:

Hi Soheil, by default vehicles do not enter an intersection if they expect to stop inside. This behaviour can be configured in the junction model (see parameter jmIgnoreKeepClearTime). Please note that junction model parameters are linked to

ZjQcmQRYFpfptBannerStart

This Message Is From an Untrusted Sender

You have not previously corresponded with this sender.

See https://itconnect.uw.edu/email-tags for additional information. Please contact the UW-IT Service Center, help@xxxxxx 206.221.5000, for assistance.

 

ZjQcmQRYFpfptBannerEnd

Hi Soheil,

 

by default vehicles do not enter an intersection if they expect to stop inside. This behaviour can be configured in the junction model (see parameter jmIgnoreKeepClearTime).

Please note that junction model parameters are linked to the vehicle but not to the junction: They are applied to all junctions. The only workaround to change the behaviour locally is changing the vehicle types locally using calibrator items:. So you would need a calibrator at every entrance/exit edge of the area you want locally different driving behaviour.

 

Best regards

Mirko

 

 

 

 

-----Original-Nachricht-----

Betreff: Re: [sumo-user] fixing bugs in a jammed simulation

Datum: 2023-09-29T22:58:34+0200

Von: "Soheil Keshavarz" <soheil99@xxxxxx>

An: "Mirko Barthauer" <m.barthauer@xxxxxxxxxxx>

 

 

 

Thanks a lot, Mirko for your reply. 

One more question. In the screenshot below you can see that in jam condition and in little junctions with alleys(roads with priority=2), vehicles do not move further and there is a gap in front of the alleys. How can I prevent this and make vehicles neglect these minor junctions, in order to better utilize main streets?

 

   

 

Regards,

Soheil

 

On Mon, Sep 25, 2023 at 4:10 AM Mirko Barthauer <m.barthauer@xxxxxxxxxxx> wrote:

Hi Soheil, you have set a quite high value for lcStrategic, meaning vehicles will care much about reaching the target turning lane early. Depending on which routes your vehicles take and how close junctions are next to each other, artefacts

ZjQcmQRYFpfptBannerStart

This Message Is From an Untrusted Sender

You have not previously corresponded with this sender.

See https://itconnect.uw.edu/email-tags for additional information. Please contact the UW-IT Service Center, help@xxxxxx 206.221.5000, for assistance.

 

ZjQcmQRYFpfptBannerEnd

Hi Soheil,

 

you have set a quite high value for lcStrategic, meaning vehicles will care much about reaching the target turning lane early. Depending on which routes your vehicles take and how close junctions are next to each other, artefacts like in your screenshots may happen because maybe all vehicles turn right/left at some point... Then there may be other isues like how connections are modeled which is impossible to tell without the scenario.

 

Regarding the first question about od2trips, I don't get the question. If you have TAZ containing multiple edges each with a probability > 0 to be chosen, then trips will not start / end necessarily on the same edge.

 

Best regards

Mirko

 

 

-----Original-Nachricht-----

Betreff: [sumo-user] fixing bugs in a jammed simulation

Datum: 2023-09-23T02:04:12+0200

Von: "Soheil Keshavarz via sumo-user" <sumo-user@xxxxxxxxxxx>

An: "sumo-user@xxxxxxxxxxx" <sumo-user@xxxxxxxxxxx>

 

 

 

Hello all,

I was hoping to get some help from you regarding different settings in my simulation of downtown Seattle, in which some links get highly jammed and many vehicles teleport.

 

1- My demand file has trips in the format below:

  <trip id="1" depart="18008.0" fromTaz="5000" toTaz="5003" type="passenger" departLane="free" departSpeed="max"/>

Since od2trip command assigns departing and arriving links to each id, I did not use it. However, like the 'different-source-sink' option in od2trip, I would like to make sure that trips with same origin and destination TAZs are not only in one link.

 

2- There are some unusual behaviors in my simulation. I would appreciate any advise to correct them. 

for example, a) in the bottom link, vehicles are not going forward in the middle lane, and vehicles that want to turn left, ONLY turn to most left lane of the adjacent link.

and b) vehicles in the south use a hight congested road with lower priority to reach the north, instead of using the link with higher priority and more lanes. Shouldn't higher priority mean a higher probability of being used in routing?

By the way, below are some hyperparameters of my sumocfg file and defintion of passenger vehicles in my simulation:

 

<vType id="passenger" vClass="passenger" accel="3.2" color="yellow" decel="3.5" length="4" maxSpeed="200" lcStrategic="2000" lcSpeedGain="5" lcCooperative="1" lcSpeedGainLookahead="5" minGap="1.5" cc1="0.9" cc2="4" cc3="-8" cc4="-0.1" cc5="0.1" cc6="11.44" cc7="0.25" cc8="3.5" cc9="1.5"/>

 

Thank you in advance for your comments.

--

SOHEIL KESHAVARZ (he/him)

Graduate Student | Research Assistant at iUTS Lab

soheil99@xxxxxx  /  washington.edu