Different Results

[ZhaoBeilan825]

Hello, your code really helps me understand the traffic light signal control. These days I try to run your codes in 1x1 grid and 1x6 grid with the CoTV and PressLight, but why are the metric "Fuel consumption(l/100km)" results in my training quite different with your results described in the paper? I haven't do any modification in the code. Looking forward to communicating with you. Thanks very much.

[stevezp009]

Hello, could you please share your "pip list" of the "flow" environment? Following the tutorial, I have some problems replicating the experiment, thanks a lot.

[Guojyjy]

Hello, your code really helps me understand the traffic light signal control. These days I try to run your codes in 1x1 grid and 1x6 grid with the CoTV and PressLight, but why are the metric "Fuel consumption(l/100km)" results in my training quite different with your results described in the paper? I haven't do any modification in the code. Looking forward to communicating with you. Thanks very much.

The difference in "Fuel consumption (l/100km)" metrics between your training results and those in the CoTV paper is due to updates of SUMO output files: changed the unit of fuel consumption measurements fuel_abs from ml to mg.

This change affects the output files and the evaluation results obtained from CoTV/evaluation/getResults.py, but does not alter the conclusions drawn in the CoTV paper.

For more detailed information on these output parameters, please refer to the SUMO documentation:

• https://sumo.dlr.de/docs/Simulation/Output/TripInfo.html
• https://sumo.dlr.de/docs/Simulation/Output/EmissionOutput.html.

[Guojyjy]

Hello, could you please share your "pip list" of the "flow" environment? Following the tutorial, I have some problems replicating the experiment, thanks a lot.

requirements.txt for the pip list on Linux (Ubuntu 20.04.2)