Environment

class airlift.envs.airlift_env.AirliftEnv(world_generator=None, pickuponly=False, verbose=False, renderer=<airlift.envs.renderer.FlatRenderer object>)

Controls all aspects of the simulation/environment. The primary interface is the step method, which provides observations/states to the agents and steps the environment to the next state given agent actions

action_space(agent: str) → gym.spaces.space.Space

Takes in agent and returns the action space for that agent.

MUST return the same value for the same agent name

Default implementation is to return the action_spaces dict

clear_rewards_dict()

Reset the rewards dictionary

close()

This methods closes any renderer window.

property env_info

Returns environment info. Should not be called before the environment is reset.

property metrics

Returns metrics collected by the environment. Should be called after the episode is done.

observation_space(agent) → gym.spaces.space.Space

Takes in agent and returns the observation space for that agent.

MUST return the same value for the same agent name

Default implementation is to return the observation_spaces dict

render(mode='human')

Displays a rendered frame from the environment, if supported. Alternate render modes in the default environments are ‘rgb_array’ which returns a numpy array and is supported by all environments outside of classic, and ‘ansi’ which returns the strings printed (specific to classic environments).

reset(seed=None) → Dict

Resets the environment and generates a new random realization. If called without a seed, a new realization is generated.

Parameters

seed – Environment seed

Returns

A dictionary containing the initial observations for all agents. Individual observations of agents can be accessed using the ‘a_0…n’ keys.

state()

Returns the complete state of the environment.

step(actions)

Steps the environment base don the given actions and returns a new observation. Based on the new observation, the policy should create another set of actions for the next step.

Parameters

actions – Dictionary that contains the actions for all agents

Returns

obs, Dictionary that contains the observation for all agents. rewards,Dictionary that contains rewards. dones, Dictionary that indicates if an agent has completed a scenario. info, A dictionary containing a list of Warnings for each agent.