miniworld-maze 1.3.0__tar.gz → 1.3.1__tar.gz

{miniworld_maze-1.3.0 → miniworld_maze-1.3.1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.3
 Name: miniworld-maze
-Version: 1.3.0
+Version: 1.3.1
 Summary: Multi-room maze environments from the DrStrategy paper. Provides NineRooms-v0, SpiralNineRooms-v0, and TwentyFiveRooms-v0 gymnasium environments.
 Keywords: reinforcement-learning,environment,gymnasium,multi-room-maze,drstrategy,maze-navigation,partial-observability,3d-environments
 Author: Tim Joseph

{miniworld_maze-1.3.0 → miniworld_maze-1.3.1}/examples/live_control_opencv.py

@@ -215,9 +215,9 @@ class OpenCVLiveController:
         Returns:
             BGR image array for OpenCV or None if failed
         """
-        if self.current_info is not None and "FIRST_PERSON" in self.current_info:
+        if self.current_info is not None and ObservationLevel.FIRST_PERSON in self.current_info:
             # Get first person observation from info
-            obs_hwc = self.current_info["FIRST_PERSON"]
+            obs_hwc = self.current_info[ObservationLevel.FIRST_PERSON]
 
             # Convert RGB to BGR for OpenCV
             obs_bgr = cv2.cvtColor(obs_hwc, cv2.COLOR_RGB2BGR)

{miniworld_maze-1.3.0 → miniworld_maze-1.3.1}/examples/save_topdown_views.py

@@ -8,12 +8,11 @@ This script demonstrates:
 3. Executing 1000 random steps and visualizing agent trajectory on top-down view
 """
 
-from typing import List, Tuple
+from typing import List
 import numpy as np
 from PIL import Image
 import gymnasium as gym
 import matplotlib.pyplot as plt
-import matplotlib.patches as patches
 
 from miniworld_maze import ObservationLevel
 import miniworld_maze  # noqa: F401
@@ -24,10 +23,10 @@ def save_single_topdown_view(env: gym.Env) -> str:
     print("🎯 Generating single top-down full view...")
     
     # Reset environment to ensure consistent state
-    env.reset(seed=42)
+    _, info = env.reset(seed=42)
     
-    # Use render_top_view directly for full view
-    obs_image = env.unwrapped.render_top_view(POMDP=False)
+    # Get top-down full view from info dict
+    obs_image = info[ObservationLevel.TOP_DOWN_FULL]
     
     # Save the observation as PNG
     filename = "single_topdown_view.png"
@@ -89,7 +88,7 @@ def collect_trajectory_and_visualize(env: gym.Env) -> str:
     print("🔍 Collecting agent trajectory over 1000 random steps...")
     
     # Reset environment
-    obs, _ = env.reset(seed=42)
+    obs, info = env.reset(seed=42)
     
     # Get scene extent from environment using the new public method
     env_unwrapped = env.unwrapped
@@ -103,17 +102,17 @@ def collect_trajectory_and_visualize(env: gym.Env) -> str:
     
     # Execute 1000 random steps
     for step in range(1000):
-        # Get current agent position
-        agent_pos = env_unwrapped.agent.pos
-        trajectory_positions.append((agent_pos[0], agent_pos[2]))  # x, z coordinates
-        
-        # Take random action
+        # Take random action first to get info dict with position
         action = env.action_space.sample()
-        obs, reward, terminated, truncated, _ = env.step(action)
+        _, _, terminated, truncated, info = env.step(action)
+        
+        # Get current agent position from info dict
+        agent_pos = info["pos"]
+        trajectory_positions.append((agent_pos[0], agent_pos[1]))  # x, z coordinates
         
         # Reset if episode ended
         if terminated or truncated:
-            obs, _ = env.reset()
+            obs, info = env.reset()
         
         if (step + 1) % 200 == 0:
             print(f"   Completed {step + 1}/1000 steps")
@@ -122,7 +121,7 @@ def collect_trajectory_and_visualize(env: gym.Env) -> str:
     
     # Get top-down view for background
     env.reset(seed=42)
-    background_image = env_unwrapped.render_top_view(POMDP=False)
+    background_image = env.unwrapped.render_top_view(POMDP=False)
     
     # Create matplotlib figure
     fig, ax = plt.subplots(1, 1, figsize=(12, 12))
@@ -175,12 +174,13 @@ def main() -> None:
     print("🖼️  Top-Down View Generator")
     print("=" * 50)
     
-    # Create a single environment instance
+    # Create a single environment instance with info_obs for top-down full view
     env = gym.make(
         "NineRooms-v0",
         obs_level=ObservationLevel.TOP_DOWN_PARTIAL,
         obs_width=256,
         obs_height=256,
+        info_obs=[ObservationLevel.TOP_DOWN_FULL],
     )
     
     try:

{miniworld_maze-1.3.0 → miniworld_maze-1.3.1}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "uv_build"
 
 [project]
 name = "miniworld-maze"
-version = "1.3.0"
+version = "1.3.1"
 description = "Multi-room maze environments from the DrStrategy paper. Provides NineRooms-v0, SpiralNineRooms-v0, and TwentyFiveRooms-v0 gymnasium environments."
 readme = "README.md"
 license = {file = "LICENSE"}

{miniworld_maze-1.3.0 → miniworld_maze-1.3.1}/src/miniworld_maze/core/miniworld_gymnasium/unified_env.py

@@ -349,8 +349,8 @@ class UnifiedMiniWorldEnv(gym.Env):
             for obs_level in self.info_obs:
                 # Generate observation with the specified level
                 info_obs = self._generate_observation(observation_level=obs_level)
-                # Use the observation level name as key
-                info[str(obs_level)] = info_obs
+                # Use the observation level enum as key
+                info[obs_level] = info_obs
 
         # Return first observation with info dict for Gymnasium compatibility
         return obs, info
@@ -584,19 +584,14 @@ class UnifiedMiniWorldEnv(gym.Env):
 
     def _calculate_step_results(self, observation):
         """Calculate reward, termination, and info for step."""
-        # Generate topdown view for info if needed
-        topdown = None
-        if self.obs_level != 2:  # Not TOP_DOWN_FULL
-            topdown = self.render_top_view(POMDP=False, frame_buffer=self.topdown_fb)
-
         # Generate additional observations for info dictionary if specified
         info = {}
         if self.info_obs is not None:
             for obs_level in self.info_obs:
                 # Generate observation with the specified level
                 info_obs = self._generate_observation(observation_level=obs_level)
-                # Use the observation level name as key
-                info[str(obs_level)] = info_obs
+                # Use the observation level enum as key
+                info[obs_level] = info_obs
 
         # Check termination
         if self.step_count >= self.max_episode_steps:
@@ -604,8 +599,7 @@ class UnifiedMiniWorldEnv(gym.Env):
             reward = 0
             info.update(
                 {
-                    "pos": self.agent.pos,
-                    "mdp_view": topdown if topdown is not None else observation,
+                    "pos": np.array([self.agent.pos[0], self.agent.pos[2]]),
                 }
             )
         else:
@@ -613,8 +607,7 @@ class UnifiedMiniWorldEnv(gym.Env):
             terminated = False
             info.update(
                 {
-                    "pos": self.agent.pos,
-                    "mdp_view": topdown if topdown is not None else observation,
+                    "pos": np.array([self.agent.pos[0], self.agent.pos[2]]),
                 }
             )