Traffic-Control-Env / openenv.yaml
dootisaha25's picture
Initial OpenEnv traffic signal environment
852e969
Raw
History Blame Contribute Delete
1.35 kB
name: indian-traffic-signal-openenv
version: "1.0.0"
description: Seedable Indian mixed-traffic signal-control environment for RL agents.
# Public action names accepted by POST /step.
actions:
- NS_GREEN
- EW_GREEN
- LEFT_PRIORITY
- PEDESTRIAN_CROSS
- EXTEND_GREEN
- EMERGENCY_OVERRIDE
- ALL_RED
observations:
tick: int
lane_queues: dict
lane_waiting_time: dict
current_signal_phase: str
time_since_last_phase_switch: int
pedestrian_count: int
pedestrian_wait_time: float
emergency_vehicle: dict
rain_level: float
random_traffic_inflow: dict
tasks:
- id: single_intersection
difficulty: easy
description: Optimize a mixed-traffic four-way Indian urban intersection.
- id: rush_hour
difficulty: medium
description: Handle asymmetric commuter inflow with rain and heavy two-wheeler traffic.
- id: emergency_priority
difficulty: hard
description: Prioritize ambulances and fire trucks without letting the rest of the junction fail.
api:
endpoints:
reset: /reset
step: /step
state: /state
tasks: /tasks
grader: /grader
baseline: /baseline
entrypoint: api:app
observation_model: TrafficState
action_model: TrafficAction
reward_range: [0.0, 1.0]
deterministic: true
seed_field: seed
tags:
- openenv
- reinforcement-learning
- traffic-control
- india