When adding new Inflows with the probability param=1/12 veh/s, as opposed to vehs_per_hour=300 (note that probability=1/12 will result in roughly 3600s/12veh=300 vehicles per hour), training rllab/green_wave.py appears to learn nothing at all for ~30 iterations. This is compared to vehs_per_hour which learns almost linearly (albeit slowly) in the same amount of time. I'm wondering if this is an Inflows problem on our end or with Sumo
When adding new Inflows with the probability param=1/12 veh/s, as opposed to vehs_per_hour=300 (note that probability=1/12 will result in roughly 3600s/12veh=300 vehicles per hour), training rllab/green_wave.py appears to learn nothing at all for ~30 iterations. This is compared to vehs_per_hour which learns almost linearly (albeit slowly) in the same amount of time. I'm wondering if this is an Inflows problem on our end or with Sumo