Factoring wind bias into simulation initialization

pyrorl/pyrorl/envs/environment/environment.py

@@ -12,7 +12,7 @@
 from scipy.stats import bernoulli
 import torch
 
-from .environment_constant import fire_mask
+from .environment_constant import fire_mask, linear_wind_transform
 
 """
 Indices corresponding to each layer of state
@@ -29,7 +29,18 @@ class FireWorld:
     while the 5 represents each of the following: [fire, fuel, populated_areas, evacuating, paths]
     """
 
-    def __init__(self, num_rows, num_cols, populated_areas, paths, paths_to_pops, num_fire_cells = 2, custom_fire_locations = None):
+    def __init__(
+            self,
+            num_rows,
+            num_cols,
+            populated_areas,
+            paths,
+            paths_to_pops,
+            num_fire_cells = 2,
+            custom_fire_locations = None,
+            wind_speed = None,
+            wind_angle = None,
+        ):
         """
         The constructor defines the state and action space, initializes the fires,
         and sets the paths and populated areas.
@@ -93,6 +104,13 @@ def __init__(self, num_rows, num_cols, populated_areas, paths, paths_to_pops, nu
         # Set the timestep
         self.time_step = 0
 
+        #Factor in wind speeds
+        if wind_speed is not None or wind_angle is not None:
+            if wind_speed is None or wind_angle is None:
+                raise TypeError("When setting wind details, wind speed and wind angle must both be provided")
+            global fire_mask
+            fire_mask = linear_wind_transform(wind_speed, wind_angle)
+
     def sample_fire_propogation(self):
         """
         Sample the next state of the wildfire model.

pyrorl/pyrorl/envs/environment/environment_constant.py

@@ -34,4 +34,4 @@
 def linear_wind_transform(wind_speed : float, wind_angle : float):
     wind_vector = np.array([[np.cos(wind_angle)], [np.sin(wind_angle)]])
     scaling_term = -(neighbor_vectors @ wind_vector) * speed_to_percent_ratio * wind_speed + 1
-    return np.clip(scaling_term * base_fire_mask, a_min=0, a_max=1)
+    return np.clip(torch.from_numpy(scaling_term) * base_fire_mask, a_min=0, a_max=1)

pyrorl/pyrorl/envs/pyrorl.py

@@ -19,7 +19,16 @@
 GRASS_COLOR = pygame.Color("#06d6a0")
 
 class WildfireEvacuationEnv(gym.Env):
-    def __init__(self, num_rows, num_cols, populated_areas, paths, paths_to_pops):
+    def __init__(
+        self,
+        num_rows,
+        num_cols,
+        populated_areas,
+        paths,
+        paths_to_pops,
+        wind_speed = None,
+        wind_angle = None,
+    ):
         """
         Set up the basic environment and its parameters.
         """
@@ -29,7 +38,15 @@ def __init__(self, num_rows, num_cols, populated_areas, paths, paths_to_pops):
         self.populated_areas = populated_areas
         self.paths = paths
         self.paths_to_pops = paths_to_pops
-        self.fire_env = FireWorld(num_rows, num_cols, populated_areas, paths, paths_to_pops)
+        self.fire_env = FireWorld(
+            num_rows,
+            num_cols,
+            populated_areas,
+            paths,
+            paths_to_pops,
+            wind_speed=wind_speed,
+            wind_angle=wind_angle
+        )
 
         # Set up action space
         actions = self.fire_env.get_actions()