def Step(self, action_request, context):
observation, reward, done, _ = self.env.step(action_request.value)
observation_pb = gym_uds_pb2.Observation(data=observation.ravel(),
shape=observation.shape)
return gym_uds_pb2.State(observation=observation_pb,
reward=reward,
done=done)
def Reset(self, empty_request, context):
observation = self.env.reset()
observation_pb = gym_uds_pb2.Observation(data=observation.ravel(),
shape=observation.shape)
return gym_uds_pb2.State(observation=observation_pb,
reward=0.0,
done=False)
def Step(self, action_request, context):
act = self.PrepareAction(action_request.data)
obs, reward, done, _ = self.env.step(act)
observation = self.PrepareObservation(obs)
assert type(observation) is np.ndarray
observation_pb = gym_uds_pb2.Observation(data=observation.ravel(), shape=observation.shape)
return gym_uds_pb2.State(observation=observation_pb, reward=reward, done=done, envID = self.env_id)
def reset(self):
observation = self.env.reset()
observation_pb = gym_uds_pb2.Observation(data=observation.ravel(),
shape=observation.shape)
utils.send_message(
self.sock,
gym_uds_pb2.State(observation=observation_pb,
reward=0.0,
done=False))
def Reset(self, empty_request, context):
observation = self.env.reset()
if (isinstance(observation, (np.ndarray, np.generic))):
observation_pb = gym_uds_pb2.Observation(data=observation.ravel(), shape=observation.shape)
elif (isinstance(observation, tuple)):
obs = np.array(observation)
observation_pb = gym_uds_pb2.Observation(data=obs.ravel(), shape=obs.shape)
else:
obs = np.array(float(observation))
observation_pb = gym_uds_pb2.Observation(data=obs.ravel(), shape=obs.shape)
return gym_uds_pb2.State(observation=observation_pb, reward=0.0, done=False, envID = self.env_id)
def step(self):
action = utils.recv_message(self.sock, gym_uds_pb2.Action)
observation, reward, done, _ = self.env.step(action.value)
assert type(observation) is np.ndarray
observation_pb = gym_uds_pb2.Observation(data=observation.ravel(),
shape=observation.shape)
utils.send_message(
self.sock,
gym_uds_pb2.State(observation=observation_pb,
reward=reward,
done=done))