habitat.core.embodied_task.Action class

An action that can be performed by an agent solving a task in environment. For example for navigation task action classes will be: MoveForwardAction, TurnLeftAction, TurnRightAction. The action can use Task members to pass a state to another action, as well as keep own state and reset when new episode starts.

Methods

def reset(self, *args: typing.Any, **kwargs: typing.Any) -> None
Reset method is called from Env on each reset for each new episode. Goal of the method is to reset Action‘s state for each episode.
def step(self, *args: typing.Any, **kwargs: typing.Any) -> simulator.Observations
Step method is called from Env on each step. Can call simulator or task method, change task’s state.

Special methods

def __init__(self, *args: typing.Any, **kwargs: typing.Any) -> None

Properties

action_space: gym.spaces.space.Space get
a current Action’s action space.

Method documentation

def habitat.core.embodied_task.Action.step(self, *args: typing.Any, **kwargs: typing.Any) -> simulator.Observations

Step method is called from Env on each step. Can call simulator or task method, change task’s state.

Parameters
args
kwargs optional parameters for the action, like distance/force.
Returns observations after taking action in the task, including ones coming from a simulator.