def make_world(self):
world = HazardousWorld()
# set scenario-specific world parameters
world.collaborative = True
world.systemic_rewards = True
world.identical_rewards = False
world.dim_c = 0 # observation-based communication
world.max_communication_distance = _MAX_COMMUNICATION_DISTANCE
world.distance_resistance_gain = _DISTANCE_RESISTANCE_GAIN
world.distance_resistance_exponent = _DISTANCE_RESISTANCE_EXPONENT
# add landmarks
world.origin_terminal_landmark = RiskRewardLandmark( risk_fn=None, reward_fn=RadialReward(1.0, 10.0))
world.destination_terminal_landmark = RiskRewardLandmark( risk_fn=None, reward_fn=RadialReward(1.0, 10.0))
world.landmarks = [world.origin_terminal_landmark, world.destination_terminal_landmark]
for lm in _NON_TERMINAL_LANDMARKS:
world.landmarks.append(lm)
for i, landmark in enumerate(world.landmarks):
landmark.name = 'landmark_%d' % i
landmark.collide = False
landmark.movable = False
landmark.size = _LANDMARK_SIZE
# properties for landmarks
if isinstance(landmark, RiskRewardLandmark) and landmark.is_hazard:
#TODO: make colors heatmap of risk probability over all bounds
landmark.color = np.array([landmark.risk_fn.get_failure_probability(0,0) + .1, 0, 0])
else:
landmark.color = np.array([0.25, 0.25, 0.25])
# make initial conditions
self.reset_world(world)
return world
def make_world(self):
world = HazardousWorld()
# set scenario-specific world parameters
world.collaborative = True
world.systemic_rewards = True
world.identical_rewards = False
world.dim_c = 0 # observation-based communication
world.max_communication_distance = _MAX_COMMUNICATION_DISTANCE
world.distance_resistance_gain = _DISTANCE_RESISTANCE_GAIN
# create and add terminal landmarks
# no intermediate landmarks in this scenario
world.origin_terminal_landmark = RiskRewardLandmark(
risk_fn=None, reward_fn=RadialReward(1.0, 10.0))
world.origin_terminal_landmark.name = 'origin'
world.origin_terminal_landmark.state.p_pos = np.array([-0.75, -0.75])
world.destination_terminal_landmark = RiskRewardLandmark(
risk_fn=None, reward_fn=RadialReward(1.0, 10.0))
world.destination_terminal_landmark.name = 'destination'
world.destination_terminal_landmark.state.p_pos = np.array(
[0.75, 0.75])
# create landmark list and set properties
world.landmarks = [
world.origin_terminal_landmark, world.destination_terminal_landmark
]
for i, landmark in enumerate(world.landmarks):
landmark.p_vel = np.zeros(world.dim_p)
landmark.collide = False
landmark.movable = False
landmark.size = _LANDMARK_SIZE
landmark.color = np.array(
[landmark.risk_fn.get_failure_probability(0, 0) + .1, 0, 0])
# make initial conditions
self.reset_world(world)
return world