python3 -m sample --pg-type td3
check out flow/flow/envs for detailed information
- one action for each intersection
- range within 0 and 1
- fully observable state space (TrafficLightGridEnv)
- for vehicles:
- velocity
- distance from the next intersection
- the unique edge it is traveling on
- for each traffic light:
- current state (the flowing direction)
- last changed time
- whether it's yellow
- for vehicles:
- large delay penalty
- switch penalty
01 23 45 11 10
9 8
7 6
- step function is at flow/envs/base.py
- env.k.vehicle.get_speed(veh_id)
- env.k.network.max_speed()
- env.network.node_mapping: iteratively give node and edge
- env.k.vehicle.get_edge(veh_id): current edge veh_id is at
- env.k.vehicle.get_ids_by_edge(edge): gives ids
- for k in env.k.network.get_edge_list()