[Feature] Navigation: kwargs for spawning ranges (#139)

* Introduce x_semidim and y_semidim parameters for managing environment dimensions

* [discovery]introduce x_semidim and y_semidim parameters for managing environment dimensions

* [navigation]introduce x_semidim and y_semidim parameters for managing environment dimensions

* [environment rendering] implementing boundary visualization for limited environments

* [navigation] ensure entity placement within constrained environment boundaries

* [environment rendering] implementing boundary visualization for limited environments

* [Features] world_spawning_x, world_spawning_y and enforce_bounds

* [Bug Fix] world_spawning_x, world_spawning_y and enforce_bounds

* Update vmas/scenarios/navigation.py

* Update vmas/scenarios/navigation.py

* [Bug Fix] world_spawning_x, world_spawning_y and enforce_bounds

* [Bug Fix] world_spawning_x, world_spawning_y and enforce_bounds

---------

Co-authored-by: Matteo Bettini <[email protected]>

Author: Giovannibriglia

vmas/scenarios/navigation.py

@@ -24,8 +24,15 @@ def make_world(self, batch_dim: int, device: torch.device, **kwargs):
         self.n_agents = kwargs.pop("n_agents", 4)
         self.collisions = kwargs.pop("collisions", True)
 
-        self.x_semidim = kwargs.pop("x_semidim", None)
-        self.y_semidim = kwargs.pop("y_semidim", None)
+        self.world_spawning_x = kwargs.pop(
+            "world_spawning_x", 1
+        )  # X-coordinate limit for entities spawning
+        self.world_spawning_y = kwargs.pop(
+            "world_spawning_y", 1
+        )  # Y-coordinate limit for entities spawning
+        self.enforce_bounds = kwargs.pop(
+            "enforce_bounds", False
+        )  # If False, the world is unlimited; else, constrained by world_spawning_x and world_spawning_y.
 
         self.agents_with_same_goal = kwargs.pop("agents_with_same_goal", 1)
         self.split_goals = kwargs.pop("split_goals", False)
@@ -43,9 +50,15 @@ def make_world(self, batch_dim: int, device: torch.device, **kwargs):
         ScenarioUtils.check_kwargs_consumed(kwargs)
 
         self.min_distance_between_entities = self.agent_radius * 2 + 0.05
-        self.world_semidim = 1
         self.min_collision_distance = 0.005
 
+        if self.enforce_bounds:
+            self.x_semidim = self.world_spawning_x
+            self.y_semidim = self.world_spawning_y
+        else:
+            self.x_semidim = None
+            self.y_semidim = None
+
         assert 1 <= self.agents_with_same_goal <= self.n_agents
         if self.agents_with_same_goal > 1:
             assert (
@@ -135,8 +148,8 @@ def reset_world_at(self, env_index: int = None):
             self.world,
             env_index,
             self.min_distance_between_entities,
-            (-self.world_semidim, self.world_semidim),
-            (-self.world_semidim, self.world_semidim),
+            (-self.world_spawning_x, self.world_spawning_x),
+            (-self.world_spawning_y, self.world_spawning_y),
         )
 
         occupied_positions = torch.stack(
@@ -152,8 +165,8 @@ def reset_world_at(self, env_index: int = None):
                 env_index=env_index,
                 world=self.world,
                 min_dist_between_entities=self.min_distance_between_entities,
-                x_bounds=(-self.world_semidim, self.world_semidim),
-                y_bounds=(-self.world_semidim, self.world_semidim),
+                x_bounds=(-self.world_spawning_x, self.world_spawning_x),
+                y_bounds=(-self.world_spawning_y, self.world_spawning_y),
             )
             goal_poses.append(position.squeeze(1))
             occupied_positions = torch.cat([occupied_positions, position], dim=1)