eclipse-sumo / sumo

Eclipse SUMO is an open source, highly portable, microscopic and continuous traffic simulation package designed to handle large networks. It allows for intermodal simulation including pedestrians and comes with a large set of tools for scenario creation.
https://eclipse.dev/sumo
Eclipse Public License 2.0
2.56k stars 1.43k forks source link

A question about LC2013 and DK2008 #13613

Closed Sirius-1110 closed 1 year ago

Sirius-1110 commented 1 year ago

As SUMO’s Lane-Changing Model descibed that LC2013 was developed by Jakob Erdmann based on DK2008, and DK2008 is the lane-changing model of the paper, Traffic Simulation with SUMO – Simulation of Urban Mobility. Here, I have a few questions for you:

  1. Why did you study a new lane-changing model at that time? Why not go with some classic rule-based lane change models?
  2. What are the advantages of DK2008 or LC2013 over other models? At present, it is rare to use LC2013 as a lane-changing model in academia. How to prove that LC2013 is an authoritative lane-changing model? Thanks for your help!
m-kro commented 1 year ago

Of which lane-change models do you think as possible alternatives? Could you name them and possibly give a reference?

Sirius-1110 commented 1 year ago

Of which lane-change models do you think as possible alternatives? Could you name them and possibly give a reference?

What about the Gipps model and its variants (MITSIM or CORISM)?

Most of the current micro traffic simulation software packages are based on rule-based classical models(Gipps?). However, there are generally problems of poor compatibility in the simulation software, such as the compatibility between the lane change model and the car-following model, and the compatibility between micro-cosmic lane-changing and macro-traffic flow, etc.

So, could it be considered that DK 2008 or LC2013 has overcome the above shortcomings?

Sirius-1110 commented 1 year ago

To add, do the Gipps model and its variants still have the following problems: **1. The amount of calculation is large, and it is not suitable for large-scale road network simulation;

  1. Lack of real-world data validation;**

Therefore, has LC2013 (DK2008) taken targeted measures to address the above problems?

behrisch commented 1 year ago

I think the best source on the topic is still Jakob's 2014 paper. So yes we did empirical validation not so much for the lane changing process itself but for lane usage. And as far as I remember the different motivations we use in SUMO for lane changing (tactical, cooperation, spped gain, keep-right) were not modelled properly in the existing models. Furthermore as you already noted there is always a strong interdependence to the car folllowing and also the junction model in use.

Sirius-1110 commented 1 year ago

I think the best source on the topic is still Jakob's 2014 paper. So yes we did empirical validation not so much for the lane changing process itself but for lane usage. And as far as I remember the different motivations we use in SUMO for lane changing (tactical, cooperation, spped gain, keep-right) were not modelled properly in the existing models. Furthermore as you already noted there is always a strong interdependence to the car folllowing and also the junction model in use.

Sorry for not getting back to you sooner, your kind assistance helps me a lot! And thank you very much for your contribution to the development of sumo!