flatland.trajectories.policy_runner module

class flatland.trajectories.policy_runner.PolicyRunner(policy: Policy, trajectory: Trajectory, by_pass_env: RailEnv | None = None, callbacks: FlatlandCallbacks | None = None, end_step=None, obs_builder: ObservationBuilder[Any, RailEnv] | None = None, rewards: Rewards | None = None, effects_generator: EffectsGenerator[RailEnv] | None = None)

Bases: object

change_policy(policy: Policy, obs_builder: ObservationBuilder)

static create_from_policy(policy: Policy, data_dir: Path, env: AbstractRailEnv | None = None, snapshot_interval: int = 1, ep_id: str | None = None, callbacks: FlatlandCallbacks | None = None, tqdm_kwargs: dict | None = None, end_step: int | None = None, no_save: bool = False) → Trajectory

property env_time

property policy

static resume(policy: Policy, trajectory: Trajectory, by_pass_env: RailEnv | None = None, snapshot_interval: int = 1, callbacks: FlatlandCallbacks | None = None, tqdm_kwargs: dict | None = None, end_step: int | None = None, obs_builder: ObservationBuilder[Any, RailEnv] | None = None, rewards: Rewards | None = None, effects_generator: EffectsGenerator[RailEnv] | None = None) → Trajectory

Creates trajectory by running submission (policy and obs builder).

Always backs up the actions and positions for steps executed in the tsvs. Can start from existing trajectory.

Parameters

policyPolicy

the submission’s policy

trajectory : Trajectory by_pass_env : RailEnv

Deprecated: pass env directly to avoid loading env (graph envs do not support env persistence yet).

snapshot_intervalint

interval to write pkl snapshots

callbacks: FlatlandCallbacks

callbacks to run during trajectory creation

tqdm_kwargs: dict

additional kwargs for tqdm

end_stepint

stop evaluation at intermediate step excl. Capped by env’s max_episode_steps

Returns

Trajectory

step(persist: bool = False) → Tuple[Trajectory, bool]

Execute one environment step. Returns (trajectory, done).