Skip to main content

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [List Home]
Re: [sumo-user] Gap acceptance in lane changing

Dear Jakob,

 

I did what you suggested:

 

·       Increase traffic: I had tested higher flow and did not see any changes. That was because of delayed insertion. I forced insertion by increasing the departSpeed and now I see a significant difference in timeGap values.

·       Reduce cooperation changing: I tested lcCooperative equal to zero but this was not the best option since lots of vehicles come to a standstill situation (speed=0) just before the end of the ramp which is not desired in my simulation scenario.

·       Set lcAssertive to lower values (i.e 0.5 or lower): According to the definitions, requiredGap is divided by lcAssertive. So, if lcAssertive is lower than 1, the Gap will be larger. The combination 0f max departSpeed and lcAssertive=2 led to the smallest timeGaps. However, compared to the field results (van Beinum et al, 2018), with the same flow, timeGaps are still quite large. Now I have two more questions:

1.        In lane change output file, I assume the gaps (leader and follower) do not include the subject length. I add 5m as the vehicle length in calculating the gap. Is that right?

2.        Is there any other way to make the vehicles accept smaller gaps in lane changing? I even set tau=0.5, but the merging vehicles (from the ramp) just stop at the end of the ramp and do not accept small gaps until the main flow vehicles stop for them. They are just so conservative in lane changing.

 

Bests,

Solmaz

 

Reference: van Beinum, A., Farah, H., Wegman, F., Hoogendoorn, S., 2018. Driving behaviour at motorway ramps and weaving segments based on empirical trajectory data. Transp. Res. Part C Emerg. Technol. 92, 426–441. https://doi.org/10.1016/j.trc.2018.05.018

 

From: sumo-user-bounces@xxxxxxxxxxx <sumo-user-bounces@xxxxxxxxxxx> On Behalf Of Jakob Erdmann
Sent: donderdag 2 juli 2020 09:42
To: Sumo project User discussions <sumo-user@xxxxxxxxxxx>
Subject: Re: [sumo-user] Gap acceptance in lane changing

 

In your scenario most of the changing happens from the on-ramp and is aided by cooperative lane-changing of the main flow (main-flow vehicles move away from the right lane to let ramp vehicles merge).

You will see an effect of lcAssertive (and also from tau) if you either

- increase traffic

- reduce cooperative changing

- set lcAssertive to lower values (i.e 0.5 or lower)

 

 

Am Mi., 1. Juli 2020 um 17:15 Uhr schrieb Solmaz Razmi Rad - CITG <S.RazmiRad@xxxxxxxxxx>:

Dear Jakob,

 

Attached you can find the simplified project. I had to keep the number of vehicles quite high so that every vehicle has a leader and follower gap. Otherwise, there would be no gap to compare.

As you see in the my_lane_change_file.xml for lcAssertive 3 and 30, leader and follower gaps have changed marginally but not significantly.

Running the simulation with lcAssertive=3 resulted in TimeGap ( μ= 6.27, std=3.26) whereas lcAssertive=30 resulted in TimeGap ( μ= 6.07, std= 3.35). Calculations are available in Python files LC3 and LC30.

Regarding Tau, I tested Tau=20 vs Tau=1.6 and saw no change in TimeGap values.

 

Bests,

Solmaz

 

 

From: sumo-user-bounces@xxxxxxxxxxx <sumo-user-bounces@xxxxxxxxxxx> On Behalf Of Jakob Erdmann
Sent: woensdag 1 juli 2020 15:51
To: Sumo project User discussions <sumo-user@xxxxxxxxxxx>
Subject: Re: [sumo-user] Gap acceptance in lane changing

 

In my experiments there is a huge difference. Please try to find a minimal version of your scenario that reproduces your problem and then send it to the list or attach it to a github-issue.

 

Am Di., 30. Juni 2020 um 17:20 Uhr schrieb Solmaz Razmi Rad - CITG <S.RazmiRad@xxxxxxxxxx>:

Yes, I used the actual gap values (leaderGap, followerGap). Just made a mistake to mention secureGap earlier. But still, I don’t see any differences in the actual gap values (leaderGap, followerGap) with changing lcAssertive.

 

From: sumo-user-bounces@xxxxxxxxxxx <sumo-user-bounces@xxxxxxxxxxx> On Behalf Of Jakob Erdmann
Sent: dinsdag 30 juni 2020 17:12
To: Sumo project User discussions <
sumo-user@xxxxxxxxxxx>
Subject: Re: [sumo-user] Gap acceptance in lane changing

 

The secureGaps are not only influenced by tau but also by vehicle speed. Changing tau can have a major influence on scenario dynamics including  average speeds so a naive interpretation of the distribution is misleading.

 

the *secureGap values in the lanechange-output are always those for lcAssertive 1 (because when the gaps are reduced due to lcAssertive this is not really secure anymore). What you would rather look at are the actual gap values (leaderGap, followerGap) which are influenced by lcAssertive.

As above, due to the influence of lcAssertive on scenario dynamics, the distribution of gaps is hard to interpret.

You could try to look at time-headways instead i.e. leaderGap/(leaderSpeed-speed)  or leaderGap/speed (also for the secureGaps).

 

regards,

Jakob

 

 

Am Di., 30. Juni 2020 um 14:52 Uhr schrieb Solmaz Razmi Rad - CITG <S.RazmiRad@xxxxxxxxxx>:

Hi Jakob,

 

Following our discussion below, I tried to run the simulation with different lcAssertive and tau values to see how they affect the value of requiredGap (in seconds) which is calculated as: “(secureBackGap + followerMinGap + subjectLength + subjectMinGap + secureFrontGap)/Speed” and derieved from the lane_change_output file.

 

To my surprise:

·       Tau did not influence the requiredGap at all. I tested tau=1.6 s (recommended for IDM) and tau=0.5 s. Both of them resulted in the same mean and standard deviation for requiredGap.

·       As you confirmed earlier, the requiredGap is divided by the value of lcAssertive. But the results did not support this one either. Running the simulation with lcAssertive=3 resulted in requiredGap ( μ= 6.0509, std=3.239) whereas lcAssertive=30 resulted in requiredGap ( μ= 6.107, std= 3.22). It seemed like the accepted gap is adjusted by some sort of a function of lcAssertive instead of a simple division and even Higher lcAssertive value resulted in higher requiredGap!

 

Am I doing some calculations wrong?

 

Regards,

Solmaz

 

 

From: sumo-user-bounces@xxxxxxxxxxx <sumo-user-bounces@xxxxxxxxxxx> On Behalf Of Jakob Erdmann
Sent: vrijdag 19 juni 2020 15:18
To: Sumo project User discussions <
sumo-user@xxxxxxxxxxx>
Subject: Re: [sumo-user] Gap acceptance in lane changing

 

1) Yes

2) you can set minGap=0 to lower the size of accepted gaps. At high speeds you will not see adverse effects but gaps will probably appear too small at lower speeds. Some car-following models such as IDM require minGap as an additional buffer for collision-free driving.

 

Am Fr., 19. Juni 2020 um 11:55 Uhr schrieb Solmaz Razmi Rad - CITG <S.RazmiRad@xxxxxxxxxx>:

1.        So, the requireGap, depends on the followSpeed. And followSpeed should be also influenced by “tau”. Then the requireGap is always a function of tau. Is this right? Can I say, if I have a distribution for tau, I will have a variation for gap acceptance (requireGap)?

2.        And regarding the minGap, can I assume minGap=0 if lane change happens without any jam?

 

 

From: sumo-user-bounces@xxxxxxxxxxx <sumo-user-bounces@xxxxxxxxxxx> On Behalf Of Jakob Erdmann
Sent: vrijdag 19 juni 2020 11:39
To: Sumo project User discussions <
sumo-user@xxxxxxxxxxx>
Subject: Re: [sumo-user] Gap acceptance in lane changing

 

1) yes

2) these are computed by the car-follow model functions. They generally are consistent with the followSpeed function so that neither the ego vehicle nor its follower on the target lane have to perform any extra braking after the lane change is completed.

3) if a jam develops immediately after lane changing, vehicles must still be able to maintain the configured minGap, thus the value is already considered during lane changing

 

 

 

Am Fr., 19. Juni 2020 um 11:26 Uhr schrieb Solmaz Razmi Rad - CITG <S.RazmiRad@xxxxxxxxxx>:

Hi there,

 

I am trying to investigate the impact of gap acceptance in lane changing around on-ramps on the capacity of a freeway. I want to use a distribution for the gaps accepted while lane changing. I realized that the parameter which controls the gap acceptance is “lcAssertive” (But if I understood correctly, lcAssertive can influence the size of accepted gap but it doesn’t define the gap itself) . The SUMO vehicle definitions state that "the required gap is divided by the value of lcAssertive". When searching for the definition of requiredGap, I came across this page ( http://sumo-user-mailing-list.90755.n8.nabble.com/sumo-user-lcAssertive-Behavior-td3244.html ) where Jakob explained about requiredGap:  “requiredGap = secureBackGap + followerMinGap + subjectLength + subjectMinGap + secureFrontGap”.

I have a couple of questions here:

1.        What exactly is requiredGap? Is this the same as the accepted gap in lane changing?

2.        How are secureBackGap or secureFrontGap defined? Is there any relation between tau and secureBackGap or secureFrontGap?

3.       From what I have understood, MinGap is the jam distance or the distance which vehicles keep when standstill. How does this contribute to gap acceptance in lane changing?

I appreciate your help in advance.

 

Regards,

Solmaz

 

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

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

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

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

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


Back to the top