def __init__(
self,
action_space: gym.spaces.Discrete,
observation_space: SpaceDict,
num_objects: int,
num_colors: int,
num_states: int,
object_embedding_dim: int = 8,
**kwargs,
):
super().__init__(**prepare_locals_for_super(locals()))
agent_view_x, agent_view_y, view_channels = observation_space[
"minigrid_ego_image"].shape
self.actor_critic = LinearActorCritic(
self.ac_key,
action_space=action_space,
observation_space=SpaceDict({
self.ac_key:
gym.spaces.Box(
low=np.float32(-1.0),
high=np.float32(1.0),
shape=(self.object_embedding_dim * agent_view_x *
agent_view_y * view_channels, ),
)
}),
)
self.memory_key = None
self.train()
def __init__(self, sensors: List[Sensor]) -> None:
self.sensors = OrderedDict()
spaces: OrderedDict[str, Space] = OrderedDict()
for sensor in sensors:
assert (
sensor.uuid
not in self.sensors), "'{}' is duplicated sensor uuid".format(
sensor.uuid)
self.sensors[sensor.uuid] = sensor
spaces[sensor.uuid] = sensor.observation_space
self.observation_spaces = SpaceDict(spaces=spaces)
def __init__(
self,
input_uuids: List[str],
output_uuid: str,
input_height: int,
input_width: int,
max_dets: int,
detector_spatial_res: int,
detector_thres: float,
device: Optional[torch.device] = None,
device_ids: Optional[List[torch.device]] = None,
**kwargs: Any,
):
self.input_height = input_height
self.input_width = input_width
self.max_dets = max_dets
self.detector_spatial_res = detector_spatial_res
self.detector_thres = detector_thres
self.device = torch.device("cpu") if device is None else device
self.device_ids = device_ids or cast(
List[torch.device], list(range(torch.cuda.device_count())))
self.frcnn: BatchedFasterRCNN = BatchedFasterRCNN(
thres=self.detector_thres,
maxdets=self.max_dets,
res=self.detector_spatial_res,
)
spaces: OrderedDict[str, gym.Space] = OrderedDict()
shape = (self.max_dets, self.detector_spatial_res,
self.detector_spatial_res)
spaces["frcnn_classes"] = gym.spaces.Box(
low=0, # 0 is bg
high=len(self.COCO_INSTANCE_CATEGORY_NAMES) - 1,
shape=shape,
dtype=np.int64,
)
shape = (
self.max_dets * 5,
self.detector_spatial_res,
self.detector_spatial_res,
)
spaces["frcnn_boxes"] = gym.spaces.Box(low=-np.inf,
high=np.inf,
shape=shape)
assert (
len(input_uuids) == 1
), "fasterrcnn preprocessor can only consume one observation type"
observation_space = SpaceDict(spaces=spaces)
super().__init__(**prepare_locals_for_super(locals()))
def __init__(
self,
input_uuid: str,
action_space: gym.spaces.Discrete,
observation_space: SpaceDict,
hidden_size: int = 128,
num_layers: int = 1,
rnn_type: str = "GRU",
head_type: Callable[
..., ActorCriticModel[CategoricalDistr]] = LinearActorCritic,
):
super().__init__(action_space=action_space,
observation_space=observation_space)
self.hidden_size = hidden_size
self.rnn_type = rnn_type
assert (
input_uuid in observation_space.spaces
), "LinearActorCritic expects only a single observational input."
self.input_uuid = input_uuid
box_space: gym.spaces.Box = observation_space[self.input_uuid]
assert isinstance(box_space, gym.spaces.Box), (
"RNNActorCritic requires that"
"observation space corresponding to the input uuid is a Box space."
)
assert len(box_space.shape) == 1
self.in_dim = box_space.shape[0]
self.state_encoder = RNNStateEncoder(
input_size=self.in_dim,
hidden_size=hidden_size,
num_layers=num_layers,
rnn_type=rnn_type,
trainable_masked_hidden_state=True,
)
self.head_uuid = "{}_{}".format("rnn", input_uuid)
self.ac_nonrecurrent_head: ActorCriticModel[
CategoricalDistr] = head_type(
input_uuid=self.head_uuid,
action_space=action_space,
observation_space=SpaceDict({
self.head_uuid:
gym.spaces.Box(low=np.float32(0.0),
high=np.float32(1.0),
shape=(hidden_size, ))
}),
)
self.memory_key = "rnn"
def __init__(
self,
action_space: gym.spaces.Discrete,
observation_space: SpaceDict,
num_objects: int,
num_colors: int,
num_states: int,
object_embedding_dim: int = 8,
hidden_size=512,
num_layers=1,
rnn_type="GRU",
head_type: Callable[
..., ActorCriticModel[CategoricalDistr]
] = LinearActorCritic,
**kwargs,
):
super().__init__(**prepare_locals_for_super(locals()))
self._hidden_size = hidden_size
agent_view_x, agent_view_y, view_channels = observation_space[
"minigrid_ego_image"
].shape
self.actor_critic = RNNActorCritic(
input_uuid=self.ac_key,
action_space=action_space,
observation_space=SpaceDict(
{
self.ac_key: gym.spaces.Box(
low=np.float32(-1.0),
high=np.float32(1.0),
shape=(
self.object_embedding_dim
* agent_view_x
* agent_view_y
* view_channels,
),
)
}
),
hidden_size=hidden_size,
num_layers=num_layers,
rnn_type=rnn_type,
head_type=head_type,
)
self.memory_key = "rnn"
self.train()
def __init__(self,
config: Config,
dataset: Optional[Dataset] = None) -> None:
assert config.is_frozen(), ("Freeze the config before creating the "
"environment, use config.freeze().")
self._config = config
self._dataset = dataset
self._current_episode_index = None
if self._dataset is None and config.DATASET.TYPE:
self._dataset = make_dataset(id_dataset=config.DATASET.TYPE,
config=config.DATASET)
self._episodes = self._dataset.episodes if self._dataset else []
self._current_episode = None
iter_option_dict = {
k.lower(): v
for k, v in config.ENVIRONMENT.ITERATOR_OPTIONS.items()
}
self._episode_iterator = self._dataset.get_episode_iterator(
**iter_option_dict)
# load the first scene if dataset is present
if self._dataset:
assert (len(self._dataset.episodes) >
0), "dataset should have non-empty episodes list"
self._config.defrost()
self._config.SIMULATOR.SCENE = self._dataset.episodes[0].scene_id
self._config.freeze()
self._sim = make_sim(id_sim=self._config.SIMULATOR.TYPE,
config=self._config.SIMULATOR)
self._task = make_task(
self._config.TASK.TYPE,
task_config=self._config.TASK,
sim=self._sim,
dataset=self._dataset,
)
self.observation_space = SpaceDict({
**self._sim.sensor_suite.observation_spaces.spaces,
**self._task.sensor_suite.observation_spaces.spaces,
})
self.action_space = self._sim.action_space
self._max_episode_seconds = (
self._config.ENVIRONMENT.MAX_EPISODE_SECONDS)
self._max_episode_steps = self._config.ENVIRONMENT.MAX_EPISODE_STEPS
self._elapsed_steps = 0
self._episode_start_time: Optional[float] = None
self._episode_over = False
def __init__(
self,
input_uuids: List[str],
output_uuid: str,
input_height: int,
input_width: int,
max_dets: int,
detector_spatial_res: int,
detector_thres: float,
parallel: bool = False,
device: Optional[torch.device] = None,
device_ids: Optional[List[torch.device]] = None,
**kwargs: Any,
):
self.input_height = input_height
self.input_width = input_width
self.max_dets = max_dets
self.detector_spatial_res = detector_spatial_res
self.detector_thres = detector_thres
self.parallel = parallel
self.device = (device if device is not None else (
"cuda" if self.parallel and torch.cuda.is_available() else "cpu"))
self.device_ids = device_ids or cast(
List[torch.device], list(range(torch.cuda.device_count())))
self.frcnn: Union[BatchedFasterRCNN,
torch.nn.DataParallel] = BatchedFasterRCNN(
thres=self.detector_thres,
maxdets=self.max_dets,
res=self.detector_spatial_res,
)
if self.parallel:
assert (torch.cuda.is_available()
), "attempt to parallelize detector without cuda"
get_logger().info("Distributing detector")
self.frcnn = self.frcnn.to(torch.device("cuda"))
self.frcnn = torch.nn.DataParallel(self.frcnn,
device_ids=self.device_ids)
get_logger().info("Detected {} devices".format(
torch.cuda.device_count()))
spaces: OrderedDict[str, gym.Space] = OrderedDict()
shape = (self.max_dets, self.detector_spatial_res,
self.detector_spatial_res)
spaces["frcnn_classes"] = gym.spaces.Box(
low=0, # 0 is bg
high=len(self.COCO_INSTANCE_CATEGORY_NAMES) - 1,
shape=shape,
dtype=np.int64,
)
shape = (
self.max_dets * 5,
self.detector_spatial_res,
self.detector_spatial_res,
)
spaces["frcnn_boxes"] = gym.spaces.Box(low=-np.inf,
high=np.inf,
shape=shape)
assert (
len(input_uuids) == 1
), "fasterrcnn preprocessor can only consume one observation type"
observation_space = SpaceDict(spaces=spaces)
super().__init__(**prepare_locals_for_super(locals()))
