Two logic after executing setRedYellowGreenState

[Gavin-Tao]

Hi @LucasAlegre

I am running ql_single-intersection.py and found that the number of logic after running setProgramLogic is 1 while that of logic after running setRedYellowGreenState is 2.

This is the output: c: (Logic(programID='0', type=0, currentPhaseIndex=0, phases=(Phase(duration=42.0, state='GGrr', minDur=-1.0, maxDur=-1.0), Phase(duration=42.0, state='rrGG', minDur=-1.0, maxDur=-1.0), Phase(duration=2.0, state='yyrr', minDur=-1.0, maxDur=-1.0), Phase(duration=2.0, state='rryy', minDur=-1.0, maxDur=-1.0)), subParameter={}),) b (Logic(programID='0', type=0, currentPhaseIndex=0, phases=(Phase(duration=42.0, state='GGrr', minDur=-1.0, maxDur=-1.0), Phase(duration=42.0, state='rrGG', minDur=-1.0, maxDur=-1.0), Phase(duration=2.0, state='yyrr', minDur=-1.0, maxDur=-1.0), Phase(duration=2.0, state='rryy', minDur=-1.0, maxDur=-1.0)), subParameter={}), Logic(programID='online', type=0, currentPhaseIndex=0, phases=(Phase(duration=1.0, state='GGrr', minDur=1.0, maxDur=1.0),), subParameter={}))

May I ask if the two logics may have an influence on training performance? Because I've run the codes but noticed that the performance isn't as expected.

Second question: I run the ql_single-intersection.py, ql_2way-single-intersection.py, and dqn_2way-single-intersection.py successfully but the output csv files seem to be not converged. I plotted the metrics but found that it seems that the performance is not stable and works well. May I ask if you verified the three scenarios' performance before? In addition, could you provide details on the output of the experiment for the result.png file in the outputs folder like which experiment's output?

I look forward to hearing from you. Your reply is really appreciated.

Thanks a lot!