Re: [sumo-user] Issues with Vehicle Counts Using Flowrouter in SUMO Simu

[Manohara C V <manohara.cv@xxxxxxxxxxxxxxxxxxx>]

Thank you, sir, I really appreciate this detailed response.

Best regards,

Manohar

---

From: Jakob Erdmann <namdre.sumo@xxxxxxxxx>
Sent: 30 July 2024 15:28
To: Manohara C V <manohara.cv@xxxxxxxxxxxxxxxxxxx>
Cc: Sumo project User discussions <sumo-user@xxxxxxxxxxx>
Subject: Re: [sumo-user] Issues with Vehicle Counts Using Flowrouter in SUMO Simulation

 

External: This email originated from outside of the organization. Do not click links or open attachments unless you recognize the sender and know the content is safe.

> I’m seeking guidance on which route files should be used as input for RouteSampler.

- routes created by dfrouter and flowRouter are prone to contain implausible detours and should not be used as input

- duarouter is used in the background when running randomTrips (to find shortest routes between random origin/destination pairs and to validate such pairs for connectivity)

- randomTrips is the recommended tool for generating routeSampler input.

> Could you please review these commands and advise if any corrections or adjustments are needed?

- you didn't give the command that converts morningrush.csv into edgeData.xml but I would recommend: 

  tools/detector/edgeDataFromFlow.py -d modified_detector_types.xml -f morning_rush.csv -o edgedata.xml -b 540 -e 660 -i 60

- the randomTrips call doesn't require the --trip-attributes options since the generated random vehicles are never used inside a simulation. Only the bare routes are used as routeSampler input

- your network contains many short edges at the network fringe. by setting routeSampler option -l (--length) you are making it very unlikely that traffic will start there

- my recommended call would be:

tools/randomTrips.py -n up.net.xml --seed 42 -r random_routes.xml  --vclass passenger --prefix veh --min-distance 300 -e 20000 --fringe-factor 10 

- your counting data references 20 edges that don't permit passenger traffic (but are instead for bicyles or rail_urban) this could indicate detector mapping errors

- if you want to have a more uniform distribution of traffic over time, replace "-f poisson" with "-f number" in the calls below

If you remove data for these edges, then routeSampler will achieve a good data match:

tools/routeSampler.py -r random_routes.xml --edgedata-files edgedata2.xml -o routesampler2.rou.xml --optimize full --mismatch-output mismatch2.xml -I -f poisson --edgedata-attribute qPKW

32400: Wrote 12827 routes (1950 distinct) achieving total count 32232 (99.92%) at 108 locations. GEH<5 for 99.07%
36000: Wrote 5455 routes (1101 distinct) achieving total count 13453 (100.16%) at 108 locations. GEH<5 for 99.07%
39600: Wrote 5316 routes (1137 distinct) achieving total count 13200 (98.00%) at 108 locations. GEH<5 for 98.15%

If you want to avoid short (one-detector-routes) you can call instead:

tools/routeSampler.py -r random_routes.xml --edgedata-files edgedata2.xml -o routesampler2.rou.xml --optimize full --mismatch-output mismatch2.xml -I -f poisson --edgedata-attribute qPKW --min-count 2

32400: Wrote 9150 routes (1001 distinct) achieving total count 30056 (93.17%) at 108 locations. GEH<5 for 93.52%
36000: Wrote 3843 routes (453 distinct) achieving total count 12261 (91.28%) at 108 locations. GEH<5 for 91.67%
39600: Wrote 3795 routes (569 distinct) achieving total count 12324 (91.50%) at 108 locations. GEH<5 for 91.67%

this does reduce the counting quality but is still within acceptance bounds of the DMRB https://en.wikipedia.org/wiki/GEH_statistic

regards,

Jakob

Am Di., 30. Juli 2024 um 10:18 Uhr schrieb Manohara C V <manohara.cv@xxxxxxxxxxxxxxxxxxx>:

Hi Jakob,

I hope you're doing well.

I am currently working with the RouteSampler tool and have the following files for your review:

• Network File: up.net.xml
• Demand File: morningrush.csv
• Detector File: modified_detector_types.xml

I’m seeking guidance on which route files should be used as input for RouteSampler. Specifically, I would like to understand the differences between the routes generated by the following methods:

• dfrouter
• flowrouter
• duarouter
• randomtrip

For a uniform distribution scenario, which routing method would be most appropriate? Additionally, for a real-world scenario, which routing method would be recommended?

Also, I have included a .txt file with the sequential commands I’m using. Could you please review these commands and advise if any corrections or adjustments are needed?

Thank you for your assistance.

Best regards,
Manohar

---

---

From: sumo-user <sumo-user-bounces@xxxxxxxxxxx> on behalf of Jakob Erdmann via sumo-user <sumo-user@xxxxxxxxxxx>
Sent: 29 July 2024 19:35
To: Sumo project User discussions <sumo-user@xxxxxxxxxxx>
Cc: Jakob Erdmann <namdre.sumo@xxxxxxxxx>
Subject: Re: [sumo-user] Issues with Vehicle Counts Using Flowrouter in SUMO Simulation

 

External: This email originated from outside of the organization. Do not click links or open attachments unless you recognize the sender and know the content is safe.

1 & 2: help on this matter requires a data sample to replicate the problem you are facing

3 & 4: my recommendation is the routeSampler tool. Of all the available count-to-traffic tools in SUMO, it's also the one that receives the most ongoing development effort. If you cannot get it to work for you, provide a data sample (network and counting data input) and I'll look into the matter myself.

regards,

Jakob

PS: response will likely be delayed due to summer holidays

Am Mo., 29. Juli 2024 um 14:07 Uhr schrieb Manohara C V via sumo-user <sumo-user@xxxxxxxxxxx>:

Dear SUMO Community,

I hope this message finds you well. I have been working on a SUMO simulation project for some time, where I am trying to simulate vehicle movements based on real-world data collected from smart cameras at various counting points.

In the process of creating these simulations, I have referred to the official documentation multiple times, exploring various tools such as flowrouter, dfrouter, jtrouter, and others. Despite my efforts, I am encountering an issue with the flowrouter tool, where it appears to generate fewer routes and vehicles than indicated in the CSV file containing the real-world counts.

Here are the specific points of concern:

1. Discrepancy in Vehicle Counts: The number of vehicles generated by flowrouter does not match the vehicle counts provided in the CSV file. This discrepancy is impacting the accuracy and reliability of my simulations.
2. Configuration and Parameters: I have meticulously followed the configuration guidelines provided in the documentation, yet the output remains inconsistent. I suspect there might be specific parameters or settings that I may have overlooked or misconfigured.
3. Alternative Tools and Methods: While I have also experimented with dfrouter, jtrouter, and other tools, I have not been able to achieve the desired results. Each tool presents its own set of challenges and limitations in terms of accurately reflecting real-world vehicle counts.
4. Best Practices and Recommendations: I would greatly appreciate any advice or recommendations from the community on how to address these issues. Are there any best practices or additional tools that I should consider? Additionally, if anyone has encountered similar challenges, I would be grateful to learn how you resolved them.
5. Examples and Case Studies: If possible, sharing examples or case studies where accurate vehicle count simulation was successfully achieved using SUMO would be immensely helpful. Practical insights into how these challenges were overcome could provide valuable guidance.

Thank you for your time and assistance. I look forward to any insights or suggestions you may have that could help improve the accuracy of my SUMO simulations.

Best regards,

Manohara c v

_______________________________________________
sumo-user mailing list
sumo-user@xxxxxxxxxxx
To unsubscribe from this list, visit https://www.eclipse.org/mailman/listinfo/sumo-user