initial commit

docker_build.sh

@@ -1 +1 @@
-sudo docker build -t tempest_in_action .
+sudo docker build -t cacm_shielding_framework .

docker_run_experiment.sh

@@ -0,0 +1,22 @@
+DOCKER_IMAGE=gsw_ai_lab
+IMAGE_VERSION=latest
+CONTAINER_NAME=tempestpy_jupyter
+
+NOTEBOOK_DIR=./notebooks
+CONTAINER_NOTEBOOK_DIR=/opt/notebooks
+LOGRESULTS_DIR=./logresults
+CONTAINER_LOGRESULTS_DIR=/opt/logresults
+MINIGRID_DIR=./notebooks/environments/Minigrid
+CONTAINER_MINIGRID_DIR=/opt/Minigrid
+
+sudo docker run \
+--name "$CONTAINER_NAME" \
+--mount "type=bind,src=$NOTEBOOK_DIR,dst=$CONTAINER_NOTEBOOK_DIR" \
+--mount "type=bind,src=$LOGRESULTS_DIR,dst=$CONTAINER_LOGRESULTS_DIR" \
+--mount "type=bind,src=$MINIGRID_DIR,dst=$CONTAINER_MINIGRID_DIR" \
+--rm \
+-p "6006:6006" \
+-p "8888:8888" \
+--shm-size=4.86gb \
+--entrypoint sh \
+"$DOCKER_IMAGE:$IMAGE_VERSION"

docker_run_jupyter.sh

@@ -1,4 +1,4 @@
-DOCKER_IMAGE=tempest_in_action
+DOCKER_IMAGE=gsw_ai_lab
 IMAGE_VERSION=latest
 CONTAINER_NAME=tempestpy_jupyter
 

dockerfile

@@ -27,6 +27,7 @@ RUN cmake .. -DCMAKE_BUILD_TYPE=$build_type
 
 # Manage python packages
 WORKDIR /opt/
+RUN apt-get update --fix-missing
 RUN apt-get install -y python3-venv
 
 
@@ -69,26 +70,24 @@ RUN make -j $no_threads
 RUN apt-get update && apt-get install ffmpeg libsm6 libxext6 -y
 
 WORKDIR /opt/tempestpy
-RUN pip install -U "ray[all]==2.6.3"
 RUN pip install dm-tree
 RUN pip install opencv-python
 RUN pip install scikit-image
 RUN pip install torch
 RUN pip install tensorboard
 RUN pip install tensorboardX
-RUN pip install gymnasium
 RUN pip install tensorflow
 RUN pip install jupyterlab
 RUN pip install astar
 RUN pip install ipywidgets
-RUN pip install numpy
 RUN pip install matplotlib
 RUN pip install sb3-contrib
 RUN pip install opencv-python
 RUN pip install moviepy
+RUN pip install gymnasium==0.29.0
+RUN pip install numpy==1.24.4
 
 ENV M2P_BINARY=/opt/Minigrid2PRISM/build/main
+RUN apt-get install bash -y
 
-# We mount the Minigrid directory here manually again, to be able to have the directory hosted in the jupyter notebook
+ENTRYPOINT ["/bin/bash"]
-CMD ["pip", "install", "-e", "/opt/Minigrid"]
-ENTRYPOINT ["jupyter", "lab","--ip=0.0.0.0","--allow-root", "--notebook-dir=/opt/notebooks"]

notebooks/.ipynb_checkpoints/ProbabilisticShielding-checkpoint.ipynb

@@ -0,0 +1,147 @@
+{
+ "cells": [
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## Example usage of Tempestpy"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {
+    "vscode": {
+     "languageId": "plaintext"
+    }
+   },
+   "outputs": [],
+   "source": [
+    "from sb3_contrib import MaskablePPO\n",
+    "from sb3_contrib.common.wrappers import ActionMasker\n",
+    "from stable_baselines3.common.logger import Logger, CSVOutputFormat, TensorBoardOutputFormat, HumanOutputFormat\n",
+    "\n",
+    "import gymnasium as gym\n",
+    "\n",
+    "from minigrid.core.actions import Actions\n",
+    "from minigrid.core.constants import TILE_PIXELS\n",
+    "from minigrid.wrappers import RGBImgObsWrapper, ImgObsWrapper\n",
+    "\n",
+    "import tempfile, datetime, shutil\n",
+    "\n",
+    "import time\n",
+    "import os\n",
+    "\n",
+    "from utils import MiniGridShieldHandler, create_log_dir, ShieldingConfig, MiniWrapper, expname, shield_needed, shielded_evaluation, create_shield_overlay_image\n",
+    "from sb3utils import MiniGridSbShieldingWrapper, parse_sb3_arguments, ImageRecorderCallback, InfoCallback\n",
+    "\n",
+    "import os, sys\n",
+    "from copy import deepcopy\n",
+    "\n",
+    "from PIL import Image"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {
+    "vscode": {
+     "languageId": "plaintext"
+    }
+   },
+   "outputs": [],
+   "source": [
+    "GRID_TO_PRISM_BINARY=os.getenv(\"M2P_BINARY\")\n",
+    "\n",
+    "def mask_fn(env: gym.Env):\n",
+    "    return env.create_action_mask()\n",
+    "\n",
+    "def nomask_fn(env: gym.Env):\n",
+    "    return [1.0] * 7\n",
+    "\n",
+    "def main():\n",
+    "    #env = \"MiniGrid-LavaSlipperyCliff-16x13-Slip10-Time-v0\"\n",
+    "    env = \"MiniGrid-WindyCity2-v0\"\n",
+    "\n",
+    "    formula = \"Pmax=? [G ! AgentIsOnLava]\"\n",
+    "    value_for_training = 0.99\n",
+    "    shield_comparison =  \"absolute\"\n",
+    "    shielding = ShieldingConfig.Training\n",
+    "    \n",
+    "    logger = Logger(\"/tmp\", output_formats=[HumanOutputFormat(sys.stdout)])\n",
+    "    \n",
+    "    env = gym.make(env, render_mode=\"rgb_array\")\n",
+    "    image_env = RGBImgObsWrapper(env, TILE_PIXELS)\n",
+    "    env = RGBImgObsWrapper(env, 8)\n",
+    "    env = ImgObsWrapper(env)\n",
+    "    env = MiniWrapper(env)\n",
+    "\n",
+    "    \n",
+    "    env.reset()\n",
+    "    Image.fromarray(env.render()).show()\n",
+    "    \n",
+    "    shield_handlers = dict()\n",
+    "    if shield_needed(shielding):\n",
+    "        for value in [0.9, 0.95, 0.99, 1.0]:\n",
+    "            shield_handler = MiniGridShieldHandler(GRID_TO_PRISM_BINARY, \"grid.txt\", \"grid.prism\", formula, shield_value=value, shield_comparison=shield_comparison, nocleanup=True, prism_file=None)\n",
+    "            env = MiniGridSbShieldingWrapper(env, shield_handler=shield_handler, create_shield_at_reset=False)\n",
+    "\n",
+    "\n",
+    "            shield_handlers[value] = shield_handler\n",
+    "    if shield_needed(shielding):\n",
+    "        for value in [0.9, 0.95, 0.99, 1.0]:            \n",
+    "            create_shield_overlay_image(image_env, shield_handlers[value].create_shield())\n",
+    "            print(f\"The shield for shield_value = {value}\")\n",
+    "\n",
+    "    if shielding == ShieldingConfig.Training:\n",
+    "        env = MiniGridSbShieldingWrapper(env, shield_handler=shield_handlers[value_for_training], create_shield_at_reset=False)\n",
+    "        env = ActionMasker(env, mask_fn)\n",
+    "        print(\"Training with shield:\")\n",
+    "        create_shield_overlay_image(image_env, shield_handlers[value_for_training].create_shield())\n",
+    "    elif shielding == ShieldingConfig.Disabled:\n",
+    "        env = ActionMasker(env, nomask_fn)\n",
+    "    else:\n",
+    "        assert(False) \n",
+    "    model = MaskablePPO(\"CnnPolicy\", env, verbose=1, device=\"auto\")\n",
+    "    model.set_logger(logger)\n",
+    "    steps = 200\n",
+    "\n",
+    "    model.learn(steps,callback=[InfoCallback()])\n",
+    "\n",
+    "\n",
+    "\n",
+    "if __name__ == '__main__':\n",
+    "    print(\"Starting the training\")\n",
+    "    main()"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": "Python 3 (ipykernel)",
+   "language": "python",
+   "name": "python3"
+  },
+  "language_info": {
+   "codemirror_mode": {
+    "name": "ipython",
+    "version": 3
+   },
+   "file_extension": ".py",
+   "mimetype": "text/x-python",
+   "name": "python",
+   "nbconvert_exporter": "python",
+   "pygments_lexer": "ipython3",
+   "version": "3.10.12"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 4
+}

notebooks/FaultyActions.py

@@ -1,105 +0,0 @@
-#!/usr/bin/env python
-# coding: utf-8
-
-# ## Example usage of Tempestpy
-
-# In[1]:
-
-
-from sb3_contrib import MaskablePPO
-from sb3_contrib.common.wrappers import ActionMasker
-from stable_baselines3.common.logger import Logger, CSVOutputFormat, TensorBoardOutputFormat, HumanOutputFormat
-
-import gymnasium as gym
-
-from minigrid.core.actions import Actions
-from minigrid.core.constants import TILE_PIXELS
-from minigrid.wrappers import RGBImgObsWrapper, ImgObsWrapper
-
-import tempfile, datetime, shutil
-
-import time
-import os
-
-from utils import MiniGridShieldHandler, create_log_dir, ShieldingConfig, MiniWrapper, expname, shield_needed, shielded_evaluation, create_shield_overlay_image
-from sb3utils import MiniGridSbShieldingWrapper, parse_sb3_arguments, ImageRecorderCallback, InfoCallback
-
-import os, sys
-from copy import deepcopy
-
-from PIL import Image
-
-
-# In[3]:
-
-
-GRID_TO_PRISM_BINARY=os.getenv("M2P_BINARY")
-
-def mask_fn(env: gym.Env):
-    return env.create_action_mask()
-
-def nomask_fn(env: gym.Env):
-    return [1.0] * 7
-
-def main():
-    env = "MiniGrid-LavaFaultyS15-1-v0"
-    
-    formula = "Pmax=? [G ! AgentIsOnLava]"
-    value_for_training = 0.0
-    shield_comparison =  "absolute"
-    shielding = ShieldingConfig.Training
-    
-    logger = Logger("/tmp", output_formats=[HumanOutputFormat(sys.stdout)])
-    
-    env = gym.make(env, render_mode="rgb_array")
-    image_env = RGBImgObsWrapper(env, TILE_PIXELS)
-    env = RGBImgObsWrapper(env, 8)
-    env = ImgObsWrapper(env)
-    env = MiniWrapper(env)
-
-    
-    env.reset()
-    Image.fromarray(env.render()).show()
-
-    shield_values = [0.0, 0.9, 0.99, 0.999, 1.0]
-    shield_handlers = dict()
-    if shield_needed(shielding):
-        for value in shield_values: 
-            shield_handler = MiniGridShieldHandler(GRID_TO_PRISM_BINARY, "grid.txt", "grid.prism", formula, shield_value=value, shield_comparison=shield_comparison, nocleanup=False, prism_file=None)
-            env = MiniGridSbShieldingWrapper(env, shield_handler=shield_handler, create_shield_at_reset=False)
-            shield_handlers[value] = shield_handler
-
-    if shield_needed(shielding):
-        for value in shield_values:            
-            create_shield_overlay_image(image_env, shield_handlers[value].create_shield())
-            print(f"The shield for shield_value = {value}")
-
-
-    if shielding == ShieldingConfig.Training:
-        env = MiniGridSbShieldingWrapper(env, shield_handler=shield_handlers[value_for_training], create_shield_at_reset=False)
-        env = ActionMasker(env, mask_fn)
-        print("Training with shield:")
-        create_shield_overlay_image(image_env, shield_handlers[value_for_training].create_shield())
-    elif shielding == ShieldingConfig.Disabled:
-        env = ActionMasker(env, nomask_fn)
-    else:
-        assert(False) 
-    model = MaskablePPO("CnnPolicy", env, verbose=1, device="auto")
-    model.set_logger(logger)
-    steps = 20_000
-
-    assert(False)
-    model.learn(steps,callback=[InfoCallback()])
-
-
-
-if __name__ == '__main__':
-    print("Starting the training")
-    main()
-
-
-# In[ ]:
-
-
-
-

notebooks/HelloLavaGap.py

@@ -1,110 +0,0 @@
-#!/usr/bin/env python
-# coding: utf-8
-
-# ## Example usage of Tempestpy
-
-# In[1]:
-
-
-from sb3_contrib import MaskablePPO
-from sb3_contrib.common.wrappers import ActionMasker
-from stable_baselines3.common.logger import Logger, CSVOutputFormat, TensorBoardOutputFormat, HumanOutputFormat
-
-import gymnasium as gym
-
-from minigrid.core.actions import Actions
-from minigrid.core.constants import TILE_PIXELS
-from minigrid.wrappers import RGBImgObsWrapper, ImgObsWrapper
-
-import tempfile, datetime, shutil
-
-import time
-import os
-
-from utils import MiniGridShieldHandler, create_log_dir, ShieldingConfig, MiniWrapper, expname, shield_needed, shielded_evaluation, create_shield_overlay_image
-from sb3utils import MiniGridSbShieldingWrapper, parse_sb3_arguments, ImageRecorderCallback, InfoCallback
-
-import os, sys
-from copy import deepcopy
-
-from PIL import Image
-
-
-# In[ ]:
-
-
-GRID_TO_PRISM_BINARY=os.getenv("M2P_BINARY")
-
-def mask_fn(env: gym.Env):
-    return env.create_action_mask()
-
-def nomask_fn(env: gym.Env):
-    return [1.0] * 7
-
-def main():
-    env = "MiniGrid-LavaGapS6-v0"
-
-    # TODO Change the safety specification
-    formula = "Pmax=? [G !AgentIsOnLava]"
-    value_for_training = 1.0
-    shield_comparison =  "absolute"
-    shielding = ShieldingConfig.Training
-    
-    logger = Logger("/tmp", output_formats=[HumanOutputFormat(sys.stdout)])
-    
-
-    env = gym.make(env, render_mode="rgb_array")
-    image_env = RGBImgObsWrapper(env, TILE_PIXELS)
-    env = RGBImgObsWrapper(env, 8)
-    env = ImgObsWrapper(env)
-    env = MiniWrapper(env)
-
-    
-    env.reset()
-    Image.fromarray(env.render()).show()
-    input("") 
-    
-    shield_handlers = dict()
-    if shield_needed(shielding):
-        for value in [0.0, 1.0]:
-            shield_handler = MiniGridShieldHandler(GRID_TO_PRISM_BINARY, "grid.txt", "grid.prism", formula, shield_value=value, shield_comparison=shield_comparison, nocleanup=True, prism_file=None)
-            env = MiniGridSbShieldingWrapper(env, shield_handler=shield_handler, create_shield_at_reset=False)
-            shield_handlers[value] = shield_handler
-
-    print("Symbolic Description of the Model:")
-    shield_handlers[1.0].print_symbolic_model()
-    input("")
-
-    if shield_needed(shielding):
-        for value in [1.0]:
-            create_shield_overlay_image(image_env, shield_handlers[value].create_shield())
-            print(f"The shield for shield_value = {value}")
-            input("")
-    
-    if shielding == ShieldingConfig.Training:
-        env = MiniGridSbShieldingWrapper(env, shield_handler=shield_handler, create_shield_at_reset=False)
-        env = ActionMasker(env, mask_fn)
-        print("Training with shield:")
-        create_shield_overlay_image(image_env, shield_handlers[value_for_training].create_shield())
-    elif shielding == ShieldingConfig.Disabled:
-        env = ActionMasker(env, nomask_fn)
-    else:
-        assert(False) 
-    model = MaskablePPO("CnnPolicy", env, verbose=1, device="auto")
-    model.set_logger(logger)
-    steps = 20_000
-
- 
-    model.learn(steps,callback=[InfoCallback()])
-
-
-
-if __name__ == '__main__':
-    main()
-
-
-# In[ ]:
-
-
-
-

notebooks/Playground.py

@@ -1,107 +0,0 @@
-#!/usr/bin/env python
-# coding: utf-8
-
-# ## Example usage of Tempestpy
-
-# In[1]:
-
-
-from sb3_contrib import MaskablePPO
-from sb3_contrib.common.wrappers import ActionMasker
-from stable_baselines3.common.logger import Logger, CSVOutputFormat, TensorBoardOutputFormat, HumanOutputFormat
-
-import gymnasium as gym
-
-from minigrid.core.actions import Actions
-from minigrid.core.constants import TILE_PIXELS
-from minigrid.wrappers import RGBImgObsWrapper, ImgObsWrapper
-
-import tempfile, datetime, shutil
-
-import time
-import os
-
-from utils import MiniGridShieldHandler, create_log_dir, ShieldingConfig, MiniWrapper, expname, shield_needed, shielded_evaluation, create_shield_overlay_image
-from sb3utils import MiniGridSbShieldingWrapper, parse_sb3_arguments, ImageRecorderCallback, InfoCallback
-
-import os, sys
-from copy import deepcopy
-
-from PIL import Image
-
-
-# In[ ]:
-
-
-GRID_TO_PRISM_BINARY=os.getenv("M2P_BINARY")
-import gymnasium as gym
-
-
-def mask_fn(env: gym.Env):
-    return env.create_action_mask()
-
-def nomask_fn(env: gym.Env):
-    return [1.0] * 7
-
-def main():
-    # Edit 'environments/Minigrid/minigrid/envs/Playground.py' to alter the environment
-    env = "MiniGrid-Playground-v0"
-
-    # TODO Change the safety specification
-    formula = "Pmax=? [G !AgentIsOnLava]"
-    value_for_training = 1.0
-    shield_comparison =  "absolute"
-    shielding = ShieldingConfig.Training
-    
-    logger = Logger("/tmp", output_formats=[HumanOutputFormat(sys.stdout)])
-    
-
-    env = gym.make(env, render_mode="rgb_array")
-    image_env = RGBImgObsWrapper(env, TILE_PIXELS)
-    env = RGBImgObsWrapper(env, 8)
-    env = ImgObsWrapper(env)
-    env = MiniWrapper(env)
-
-    
-    env.reset()
-    Image.fromarray(env.render()).show()
-        
-    shield_handlers = dict()
-    if shield_needed(shielding):
-        for value in [0.9, 0.99, 0.999, 0.9999, 1.0]:
-            shield_handler = MiniGridShieldHandler(GRID_TO_PRISM_BINARY, "grid.txt", "grid.prism", formula, shield_value=value, shield_comparison=shield_comparison, nocleanup=True, prism_file=None)
-            env = MiniGridSbShieldingWrapper(env, shield_handler=shield_handler, create_shield_at_reset=False)
-            create_shield_overlay_image(image_env, shield_handler.create_shield())
-            print(f"The shield for shield_value = {value}")
-
-            shield_handlers[value] = shield_handler
-
-
-    if shielding == ShieldingConfig.Training:
-        env = MiniGridSbShieldingWrapper(env, shield_handler=shield_handler, create_shield_at_reset=False)
-        env = ActionMasker(env, mask_fn)
-        print("Training with shield:")
-        create_shield_overlay_image(image_env, shield_handlers[value_for_training].create_shield())
-    elif shielding == ShieldingConfig.Disabled:
-        env = ActionMasker(env, nomask_fn)
-    else:
-        assert(False) 
-    model = MaskablePPO("CnnPolicy", env, verbose=1, device="auto")
-    model.set_logger(logger)
-    steps = 20_000
-
- 
-    model.learn(steps,callback=[InfoCallback()])
-
-
-
-if __name__ == '__main__':
-    print("Starting the training")
-    main()
-
-
-# In[ ]:
-
-
-
-

notebooks/SlipperyCliff.py

@@ -1,105 +0,0 @@
-#!/usr/bin/env python
-# coding: utf-8
-
-# ## Example usage of Tempestpy
-
-# In[1]:
-
-
-from sb3_contrib import MaskablePPO
-from sb3_contrib.common.wrappers import ActionMasker
-from stable_baselines3.common.logger import Logger, CSVOutputFormat, TensorBoardOutputFormat, HumanOutputFormat
-
-import gymnasium as gym
-
-from minigrid.core.actions import Actions
-from minigrid.core.constants import TILE_PIXELS
-from minigrid.wrappers import RGBImgObsWrapper, ImgObsWrapper
-
-import tempfile, datetime, shutil
-
-import time
-import os
-
-from utils import MiniGridShieldHandler, create_log_dir, ShieldingConfig, MiniWrapper, expname, shield_needed, shielded_evaluation, create_shield_overlay_image
-from sb3utils import MiniGridSbShieldingWrapper, parse_sb3_arguments, ImageRecorderCallback, InfoCallback
-
-import os, sys
-from copy import deepcopy
-
-from PIL import Image
-
-
-# In[3]:
-
-
-GRID_TO_PRISM_BINARY=os.getenv("M2P_BINARY")
-
-def mask_fn(env: gym.Env):
-    return env.create_action_mask()
-
-def nomask_fn(env: gym.Env):
-    return [1.0] * 7
-
-def main():
-    #env = "MiniGrid-LavaSlipperyCliff-16x13-Slip10-Time-v0"
-    env = "MiniGrid-WindyCity2-v0"
-    
-    formula = "Pmax=? [G ! AgentIsOnLava]"
-    value_for_training = 0.99
-    shield_comparison =  "absolute"
-    shielding = ShieldingConfig.Training
-    
-    logger = Logger("/tmp", output_formats=[HumanOutputFormat(sys.stdout)])
-    
-    env = gym.make(env, render_mode="rgb_array")
-    image_env = RGBImgObsWrapper(env, TILE_PIXELS)
-    env = RGBImgObsWrapper(env, 8)
-    env = ImgObsWrapper(env)
-    env = MiniWrapper(env)
-
-    
-    env.reset()
-    Image.fromarray(env.render()).show()
-    
-    shield_handlers = dict()
-    if shield_needed(shielding):
-        for value in [0.9, 0.95, 0.99, 1.0]:
-            shield_handler = MiniGridShieldHandler(GRID_TO_PRISM_BINARY, "grid.txt", "grid.prism", formula, shield_value=value, shield_comparison=shield_comparison, nocleanup=True, prism_file=None)
-            env = MiniGridSbShieldingWrapper(env, shield_handler=shield_handler, create_shield_at_reset=False)
-
-
-            shield_handlers[value] = shield_handler
-    if shield_needed(shielding):
-        for value in [0.9, 0.95, 0.99, 1.0]:            
-            create_shield_overlay_image(image_env, shield_handlers[value].create_shield())
-            print(f"The shield for shield_value = {value}")
-
-    if shielding == ShieldingConfig.Training:
-        env = MiniGridSbShieldingWrapper(env, shield_handler=shield_handlers[value_for_training], create_shield_at_reset=False)
-        env = ActionMasker(env, mask_fn)
-        print("Training with shield:")
-        create_shield_overlay_image(image_env, shield_handlers[value_for_training].create_shield())
-    elif shielding == ShieldingConfig.Disabled:
-        env = ActionMasker(env, nomask_fn)
-    else:
-        assert(False) 
-    model = MaskablePPO("CnnPolicy", env, verbose=1, device="auto")
-    model.set_logger(logger)
-    steps = 20_000
-
-    assert(False)
-    model.learn(steps,callback=[InfoCallback()])
-
-
-
-if __name__ == '__main__':
-    print("Starting the training")
-    main()
-
-
-# In[ ]:
-
-
-
-

notebooks/environments/Minigrid/minigrid/envs/.ipynb_checkpoints/windycity-checkpoint.py

@@ -0,0 +1,744 @@
+from __future__ import annotations
+from minigrid.core.grid import Grid
+from minigrid.core.mission import MissionSpace
+from minigrid.core.world_object import (
+    SlipperyEast,
+    SlipperySouth,
+    SlipperyNorth,
+    SlipperyWest,
+    SlipperyNorthEast,
+    Lava,
+    Goal
+ )
+from minigrid.envs.adversaries_base import AdversaryEnv
+from minigrid.core.tasks import FollowAgent, DoRandom, GoTo
+
+from minigrid.minigrid_env import MiniGridEnv, is_slippery
+
+import numpy as np
+from itertools import product
+
+class WindyCityEnv(MiniGridEnv):
+    def __init__(self,
+                randomize_start=True, size=10,
+                width=24,
+                height=22,
+                probability_intended=8/9,
+                probability_turn_intended=8/9,
+                obstacle_type=Lava,
+                goal_reward=1,
+                failure_penalty=-1,
+                per_step_penalty=0,
+                dense_rewards=False,
+                     **kwargs):
+
+        self.obstacle_type = obstacle_type
+        self.size = size
+        self.probability_intended = probability_intended
+        self.probability_turn_intended = probability_turn_intended
+
+        if width is not None and height is not None:
+            self.width = width
+            self.height = height
+        elif size is not None:
+            self.width = size
+            self.height = size
+        else:
+            raise ValueError(f"Please define either width and height or a size for square environments. The set values are width: {width}, height: {height}, size: {size}.")
+
+        mission_space = MissionSpace(mission_func=self._gen_mission)
+        super().__init__(
+            width=self.width,
+            height=self.height,
+            max_steps=200,
+            # Set this to True for maximum speed
+            see_through_walls=False,
+            mission_space = mission_space,
+            **kwargs
+        )
+
+        self.randomize_start = randomize_start
+        self.goal_reward = goal_reward
+        self.failure_penalty = failure_penalty
+        self.dense_rewards = dense_rewards
+        self.per_step_penalty = per_step_penalty
+
+        self.trajectory = list()
+
+    @staticmethod
+    def _gen_mission():
+        return "Finish your task while avoiding the adversaries"
+
+    def disable_random_start(self):
+        self.randomize_start = False
+
+    def place_agent(self, spawn_on_slippery=False, agent_pos=None, agent_dir=0):
+        max_tries = 10_000
+        num_tries = 0
+
+        if self.randomize_start == True:
+            while True:
+                num_tries += 1
+                if num_tries > max_tries:
+                    raise RecursionError("rejection sampling failed in place_obj")
+                x = np.random.randint(0, self.width)
+                y = np.random.randint(0, self.height)
+
+                cell = self.grid.get(*(x,y))
+                if ( cell is None or
+                    (cell.can_overlap() and
+                        not isinstance(cell, Lava) and
+                        not isinstance(cell, Goal) and
+                        (spawn_on_slippery or not is_slippery(cell)) and
+                        not (x in [7, 8, 9, 10] and y in [9, 10]))
+                    ):
+                    self.agent_pos = np.array((x, y))
+                    self.agent_dir = np.random.randint(0, 4)
+                    break
+        elif agent_dir is None:
+            self.agent_pos = np.array((1, 1))
+            self.agent_dir = 0
+        else:
+            self.agent_pos = agent_pos
+            self.agent_dir = agent_dir
+        self.trajectory.append((self.agent_pos, self.agent_dir))
+
+    def place_goal(self, goal_pos):
+        self.goal_pos = goal_pos
+        self.put_obj(Goal(), *self.goal_pos)
+
+
+    def printGrid(self, init=False):
+        grid = super().printGrid(init)
+
+        properties_str = ""
+
+        properties_str += F"ProbTurnIntended:{self.probability_turn_intended}\n"
+        properties_str += F"ProbForwardIntended:{self.probability_intended}\n"
+
+        return grid + properties_str
+
+    def step(self, action):
+        obs, reward, terminated, truncated, info = super().step(action)
+        self.trajectory.append((action, self.agent_pos, self.agent_dir))
+        if truncated and info["ran_into_lava"]:
+            print(self.trajectory)
+            print("truncated: ", info)
+            self.trajectory = list()
+        if truncated and info["reached_goal"]:
+            print("truncated: ", info)
+            self.trajectory = list()
+        elif terminated and info["ran_into_lava"]:
+            print(self.trajectory)
+            print("terminated: ", info)
+            self.trajectory = list()
+        elif terminated:
+            print("terminated: ", info)
+            self.trajectory = list()
+        elif truncated:
+            print("truncated: ", info)
+            self.trajectory = list()
+        return obs, reward - self.per_step_penalty, terminated, truncated, info
+
+    def reset(self, **kwargs) -> tuple[ObsType, dict[str, Any]]:
+        return super().reset(**kwargs)
+
+    def _place_building(self, col, row, width, height, obj_type=Lava):
+        for i in range(col, width + col):
+            self.grid.vert_wall(i, row, height, obj_type=obj_type)
+
+    def _gen_grid(self, width, height):
+        super()._gen_grid(width, height)
+        self.grid = Grid(width, height)
+
+        # Generate the surrounding walls
+        self.grid.horz_wall(0, 0)
+        self.grid.horz_wall(0, height - 1)
+        self.grid.vert_wall(0, 0)
+        self.grid.vert_wall(width - 1, 0)
+
+        for i in range(1, height - 1):
+            self.grid.horz_wall(1, i, width-2, obj_type=SlipperyNorthEast("white", probability_intended=self.probability_intended, probability_turn_intended=self.probability_turn_intended))
+
+        self._place_building(13, 1, 4, 2)
+        self.grid.vert_wall(12, 1, 2, obj_type=SlipperyNorth("white", probability_intended=self.probability_intended, probability_turn_intended=self.probability_turn_intended))
+        self.grid.horz_wall(13, 3, 4, obj_type=SlipperyEast("white", probability_intended=self.probability_intended, probability_turn_intended=self.probability_turn_intended))
+        self.grid.vert_wall(17, 1, 2, obj_type=SlipperyNorth("white", probability_intended=self.probability_intended, probability_turn_intended=self.probability_turn_intended))
+
+        self._place_building(7,  3, 3, 4)
+        self.grid.vert_wall(6, 3, 4, obj_type=SlipperyNorth("white", probability_intended=self.probability_intended, probability_turn_intended=self.probability_turn_intended))
+        self.grid.vert_wall(10, 3, 4, obj_type=SlipperyNorth("white", probability_intended=self.probability_intended, probability_turn_intended=self.probability_turn_intended))
+        self.grid.horz_wall(7, 2, 3, obj_type=SlipperyEast("white", probability_intended=self.probability_intended, probability_turn_intended=self.probability_turn_intended))
+        self.grid.horz_wall(7, 7, 3, obj_type=SlipperyEast("white", probability_intended=self.probability_intended, probability_turn_intended=self.probability_turn_intended))
+
+        self._place_building(15, 7, 6, 4)
+        self.grid.vert_wall(14, 7, 4, obj_type=SlipperyNorth("white", probability_intended=self.probability_intended, probability_turn_intended=self.probability_turn_intended))
+        self.grid.vert_wall(14, 9, 2, obj_type=Lava)
+        self.grid.vert_wall(20, 7, 4, obj_type=SlipperyNorth("white", probability_intended=self.probability_intended, probability_turn_intended=self.probability_turn_intended))
+        self.grid.vert_wall(13, 9, 2, obj_type=SlipperyNorth("white", probability_intended=self.probability_intended, probability_turn_intended=self.probability_turn_intended))
+        self.grid.horz_wall(15, 6, 5, obj_type=SlipperyEast("white", probability_intended=self.probability_intended, probability_turn_intended=self.probability_turn_intended))
+        self.grid.horz_wall(14, 11, 6, obj_type=SlipperyEast("white", probability_intended=self.probability_intended, probability_turn_intended=self.probability_turn_intended))
+
+
+        self._place_building(5, 11, 5, 6)
+        self.grid.vert_wall(4, 11, 6, obj_type=SlipperyNorth("white", probability_intended=self.probability_intended, probability_turn_intended=self.probability_turn_intended))
+        self.grid.vert_wall(10, 11, 6, obj_type=SlipperyNorth("white", probability_intended=self.probability_intended, probability_turn_intended=self.probability_turn_intended))
+        self.grid.horz_wall(5, 17, 5, obj_type=SlipperyEast("white", probability_intended=self.probability_intended, probability_turn_intended=self.probability_turn_intended))
+        self.grid.horz_wall(5, 10, 5, obj_type=SlipperyWest("white", probability_intended=self.probability_intended, probability_turn_intended=self.probability_turn_intended))
+        self.grid.horz_wall(6, 9, 4, obj_type=SlipperyWest("white", probability_intended=self.probability_intended, probability_turn_intended=self.probability_turn_intended))
+        self.grid.vert_wall(9, 7, 4, obj_type=SlipperySouth("white", probability_intended=self.probability_intended, probability_turn_intended=self.probability_turn_intended))
+
+        self._place_building(21, 13, 2, 5)
+        self.grid.vert_wall(20, 13, 5, obj_type=SlipperyNorth("white", probability_intended=self.probability_intended, probability_turn_intended=self.probability_turn_intended))
+        self.grid.horz_wall(21, 12, 2, obj_type=SlipperyEast("white", probability_intended=self.probability_intended, probability_turn_intended=self.probability_turn_intended))
+        self.grid.horz_wall(21, 18, 2, obj_type=SlipperyEast("white", probability_intended=self.probability_intended, probability_turn_intended=self.probability_turn_intended))
+
+
+
+        self.place_agent(agent_pos=np.array((1, height -2)), agent_dir=0, spawn_on_slippery=True)
+        self.place_goal(np.array((width - 2, 1)))
+        if self.dense_rewards: self.run_bfs()
+
+
+class WindyCityAdvEnv(AdversaryEnv):
+    def __init__(self,
+                randomize_start=True, size=10,
+                width=15,
+                height=15,
+                probability_intended=8/9,
+                probability_turn_intended=8/9,
+                obstacle_type=Lava,
+                goal_reward=1,
+                failure_penalty=-1,
+                per_step_penalty=0,
+                dense_rewards=False,
+                     **kwargs):
+
+        self.obstacle_type = obstacle_type
+        self.size = size
+        self.probability_intended = probability_intended
+        self.probability_turn_intended = probability_turn_intended
+
+        if width is not None and height is not None:
+            self.width = width
+            self.height = height
+        elif size is not None:
+            self.width = size
+            self.height = size
+        else:
+            raise ValueError(f"Please define either width and height or a size for square environments. The set values are width: {width}, height: {height}, size: {size}.")
+
+        super().__init__(
+            width=self.width,
+            height=self.height,
+            max_steps=200,
+            # Set this to True for maximum speed
+            see_through_walls=False,
+            **kwargs
+        )
+
+        self.randomize_start = randomize_start
+        self.goal_reward = goal_reward
+        self.failure_penalty = failure_penalty
+        self.dense_rewards = dense_rewards
+        self.per_step_penalty = per_step_penalty
+
+        self.trajectory = list()
+
+    def disable_random_start(self):
+        self.randomize_start = False
+
+    def place_agent(self, spawn_on_slippery=False, agent_pos=None, agent_dir=0):
+        max_tries = 10_000
+        num_tries = 0
+
+        if self.randomize_start == True:
+            while True:
+                num_tries += 1
+                if num_tries > max_tries:
+                    raise RecursionError("rejection sampling failed in place_obj")
+                x = np.random.randint(0, self.width)
+                y = np.random.randint(0, self.height)
+
+                cell = self.grid.get(*(x,y))
+                if ( cell is None or
+                    (cell.can_overlap() and
+                        not isinstance(cell, Lava) and
+                        not isinstance(cell, Goal) and
+                        (spawn_on_slippery or not is_slippery(cell)) and
+                        not (x in [7, 8, 9, 10] and y in [9, 10]))
+                    ):
+                    self.agent_pos = np.array((x, y))
+                    self.agent_dir = np.random.randint(0, 4)
+                    break
+        elif agent_dir is None:
+            self.agent_pos = np.array((1, 1))
+            self.agent_dir = 0
+        else:
+            self.agent_pos = agent_pos
+            self.agent_dir = agent_dir
+        self.trajectory.append((self.agent_pos, self.agent_dir))
+
+    def place_goal(self, goal_pos):
+        self.goal_pos = goal_pos
+        self.put_obj(Goal(), *self.goal_pos)
+
+
+    def printGrid(self, init=False):
+        grid = super().printGrid(init)
+
+        properties_str = ""
+
+        properties_str += F"ProbTurnIntended:{self.probability_turn_intended}\n"
+        properties_str += F"ProbForwardIntended:{self.probability_intended}\n"
+
+        return grid + properties_str
+
+    def step(self, action):
+        obs, reward, terminated, truncated, info = super().step(action)
+        self.trajectory.append((action, self.agent_pos, self.agent_dir, str(self.adversaries["blue"])))
+        if truncated and info["ran_into_lava"]:
+            print(self.trajectory)
+            print("truncated: ", info)
+            self.trajectory = list()
+        if truncated and info["reached_goal"]:
+            print("truncated: ", info)
+            self.trajectory = list()
+        elif terminated and info["ran_into_lava"]:
+            print(self.trajectory)
+            print("terminated: ", info)
+            self.trajectory = list()
+        elif terminated:
+            print("terminated: ", info)
+            self.trajectory = list()
+        elif truncated:
+            print("truncated: ", info)
+            self.trajectory = list()
+        return obs, reward - self.per_step_penalty, terminated, truncated, info
+
+    def reset(self, **kwargs) -> tuple[ObsType, dict[str, Any]]:
+        return super().reset(**kwargs)
+
+    def _place_building(self, col, row, width, height, obj_type=Lava):
+        for i in range(col, width + col):
+            self.grid.vert_wall(i, row, height, obj_type=obj_type)
+
+    def _gen_grid(self, width, height):
+        super()._gen_grid(width, height)
+        self.grid = Grid(width, height)
+
+        # Generate the surrounding walls
+        self.grid.horz_wall(0, 0)
+        self.grid.horz_wall(0, height - 1)
+        self.grid.vert_wall(0, 0)
+        self.grid.vert_wall(width - 1, 0)
+
+        for i in range(1, height - 1):
+            self.grid.horz_wall(1, i, width-2, obj_type=SlipperyNorthEast("white", probability_intended=self.probability_intended, probability_turn_intended=self.probability_turn_intended))
+
+        self._place_building(7, 1, 4, 1)
+        self.grid.vert_wall(6, 1, 1, obj_type=SlipperyNorth("white", probability_intended=self.probability_intended, probability_turn_intended=self.probability_turn_intended))
+        self.grid.horz_wall(7, 2, 4, obj_type=SlipperyEast("white", probability_intended=self.probability_intended, probability_turn_intended=self.probability_turn_intended))
+        self.grid.vert_wall(11, 1, 1, obj_type=SlipperyNorth("white", probability_intended=self.probability_intended, probability_turn_intended=self.probability_turn_intended))
+
+        self._place_building(4, 5, 2, 1)
+        self.grid.vert_wall(3, 5, 1, obj_type=SlipperyNorth("white", probability_intended=self.probability_intended, probability_turn_intended=self.probability_turn_intended))
+        self.grid.vert_wall(6, 5, 1, obj_type=SlipperyNorth("white", probability_intended=self.probability_intended, probability_turn_intended=self.probability_turn_intended))
+        self.grid.horz_wall(4, 4, 2, obj_type=SlipperyEast("white", probability_intended=self.probability_intended, probability_turn_intended=self.probability_turn_intended))
+        self.grid.horz_wall(4, 6, 2, obj_type=SlipperyEast("white", probability_intended=self.probability_intended, probability_turn_intended=self.probability_turn_intended))
+
+        self._place_building(12, 7, 2, 3)
+        self.grid.vert_wall(11, 7, 3, obj_type=SlipperyNorth("white", probability_intended=self.probability_intended, probability_turn_intended=self.probability_turn_intended))
+        self.grid.horz_wall(11, 6, 3, obj_type=SlipperyEast("white", probability_intended=self.probability_intended, probability_turn_intended=self.probability_turn_intended))
+        self.grid.horz_wall(11, 10, 3, obj_type=SlipperyEast("white", probability_intended=self.probability_intended, probability_turn_intended=self.probability_turn_intended))
+
+
+        self._place_building(4, 10, 2, 2)
+        self.grid.vert_wall(3, 10, 2, obj_type=SlipperyNorth("white", probability_intended=self.probability_intended, probability_turn_intended=self.probability_turn_intended))
+        self.grid.vert_wall(6, 10, 2, obj_type=SlipperyNorth("white", probability_intended=self.probability_intended, probability_turn_intended=self.probability_turn_intended))
+        self.grid.horz_wall(4, 12, 2, obj_type=SlipperyEast("white", probability_intended=self.probability_intended, probability_turn_intended=self.probability_turn_intended))
+        self.grid.horz_wall(4, 9, 2, obj_type=SlipperyWest("white", probability_intended=self.probability_intended, probability_turn_intended=self.probability_turn_intended))
+        self.grid.vert_wall(5, 7, 3, obj_type=SlipperySouth("white", probability_intended=self.probability_intended, probability_turn_intended=self.probability_turn_intended))
+
+
+        #should spawn randomly
+        x = np.random.choice([1,2,3,6,7,8,9])
+        y = np.random.choice([6,7,8])
+        self.add_adversary(x, y, "blue", direction=1, tasks=[FollowAgent("red", duration=2), DoRandom(duration=2)], repeating=True)
+
+        self.place_agent(agent_pos=np.array((1, height -2)), agent_dir=0, spawn_on_slippery=True)
+        self.place_goal(np.array((width - 2, 1)))
+        if self.dense_rewards: self.run_bfs()
+
+class WindyCity2Env(MiniGridEnv):
+    def __init__(self,
+                randomize_start=True, size=10,
+                width=27,
+                height=22,
+                probability_intended=8/9,
+                probability_turn_intended=8/9,
+                obstacle_type=Lava,
+                goal_reward=1,
+                failure_penalty=-1,
+                per_step_penalty=0,
+                dense_rewards=False,
+                two_player_winning_region_start=False,
+                     **kwargs):
+
+        self.obstacle_type = obstacle_type
+        self.size = size
+        self.probability_intended = probability_intended
+        self.probability_turn_intended = probability_turn_intended
+
+        if width is not None and height is not None:
+            self.width = width
+            self.height = height
+        elif size is not None:
+            self.width = size
+            self.height = size
+        else:
+            raise ValueError(f"Please define either width and height or a size for square environments. The set values are width: {width}, height: {height}, size: {size}.")
+
+        mission_space = MissionSpace(mission_func=self._gen_mission)
+        super().__init__(
+            width=self.width,
+            height=self.height,
+            max_steps=200,
+            # Set this to True for maximum speed
+            see_through_walls=False,
+            mission_space = mission_space,
+            **kwargs
+        )
+
+        self.randomize_start = randomize_start
+        self.two_player_winning_region_start = two_player_winning_region_start
+        self.goal_reward = goal_reward
+        self.failure_penalty = failure_penalty
+        self.dense_rewards = dense_rewards
+        self.per_step_penalty = per_step_penalty
+
+        self.trajectory = list()
+
+    @staticmethod
+    def _gen_mission():
+        return "Finish your task while avoiding the adversaries"
+
+    def disable_random_start(self):
+        self.randomize_start = False
+
+    def place_agent(self, spawn_on_slippery=False, agent_pos=None, agent_dir=0):
+        max_tries = 10_000
+        num_tries = 0
+
+        if self.two_player_winning_region_start == True:
+            winning_region = list()
+            winning_region += product([1,2,3,4], [y for y in range(1, self.height-1)])
+            winning_region += product([x for x in range(1,12)], [1])
+            winning_region += product([x for x in range(1,self.width-10)], [self.height-2])
+            winning_region += product([x for x in range(self.width-6, self.width-1)], [1,2,3,4])
+            winning_region += product([x for x in range(self.width-11, self.width-1)], [5])
+            x, y= winning_region[np.random.choice(len(winning_region), 1)[0]]
+            self.agent_pos = np.array((x,y))
+            self.agent_dir = np.random.randint(0, 4)
+            self.trajectory.append((self.agent_pos, self.agent_dir))
+            return
+
+        if self.randomize_start == True:
+            while True:
+                num_tries += 1
+                if num_tries > max_tries:
+                    raise RecursionError("rejection sampling failed in place_obj")
+                x = np.random.randint(0, self.width)
+                y = np.random.randint(0, self.height)
+
+                cell = self.grid.get(*(x,y))
+                if ( cell is None or
+                    (cell.can_overlap() and
+                        not isinstance(cell, Lava) and
+                        not isinstance(cell, Goal) and
+                        (spawn_on_slippery or not is_slippery(cell)) and
+                        not (x in [7, 8, 9, 10] and y in [9, 10]))
+                    ):
+                    self.agent_pos = np.array((x, y))
+                    self.agent_dir = np.random.randint(0, 4)
+                    break
+        elif agent_dir is None:
+            self.agent_pos = np.array((1, 1))
+            self.agent_dir = 0
+        else:
+            self.agent_pos = agent_pos
+            self.agent_dir = agent_dir
+        self.trajectory.append((self.agent_pos, self.agent_dir))
+
+    def place_goal(self, goal_pos):
+        self.goal_pos = goal_pos
+        self.put_obj(Goal(), *self.goal_pos)
+
+
+    def printGrid(self, init=False):
+        grid = super().printGrid(init)
+
+        properties_str = ""
+
+        properties_str += F"ProbTurnIntended:{self.probability_turn_intended}\n"
+        properties_str += F"ProbForwardIntended:{self.probability_intended}\n"
+
+        return grid + properties_str
+
+    def step(self, action):
+        obs, reward, terminated, truncated, info = super().step(action)
+        self.trajectory.append((action, self.agent_pos, self.agent_dir))
+        if truncated and info["ran_into_lava"]:
+            print(self.trajectory)
+            print("truncated: ", info)
+            self.trajectory = list()
+        if truncated and info["reached_goal"]:
+            print("truncated: ", info)
+            self.trajectory = list()
+        elif terminated and info["ran_into_lava"]:
+            print(self.trajectory)
+            print("terminated: ", info)
+            self.trajectory = list()
+        elif terminated:
+            print("terminated: ", info)
+            self.trajectory = list()
+        elif truncated:
+            print("truncated: ", info)
+            self.trajectory = list()
+        return obs, reward - self.per_step_penalty, terminated, truncated, info
+
+    def reset(self, **kwargs) -> tuple[ObsType, dict[str, Any]]:
+        return super().reset(**kwargs)
+
+    def _place_building(self, col, row, width, height, obj_type=Lava):
+        for i in range(col, width + col):
+            self.grid.vert_wall(i, row, height, obj_type=obj_type)
+
+    def _gen_grid(self, width, height):
+        super()._gen_grid(width, height)
+        self.grid = Grid(width, height)
+
+        # Generate the surrounding walls
+        self.grid.horz_wall(0, 0)
+        self.grid.horz_wall(0, height - 1)
+        self.grid.vert_wall(0, 0)
+        self.grid.vert_wall(width - 1, 0)
+
+        for i in range(1, height - 1):
+            self.grid.horz_wall(1, i, width-2, obj_type=SlipperyNorthEast("white", probability_intended=self.probability_intended, probability_turn_intended=self.probability_turn_intended))
+
+        self.grid.horz_wall(1, 17, 15, obj_type=SlipperyEast("white", probability_intended=self.probability_intended, probability_turn_intended=self.probability_turn_intended))
+        self.grid.horz_wall(1, 18, 16, obj_type=SlipperyEast("white", probability_intended=self.probability_intended, probability_turn_intended=self.probability_turn_intended))
+        self.grid.horz_wall(1, 19, 17, obj_type=SlipperyEast("white", probability_intended=self.probability_intended, probability_turn_intended=self.probability_turn_intended))
+        self.grid.horz_wall(1, 20, 18, obj_type=SlipperyEast("white", probability_intended=self.probability_intended, probability_turn_intended=self.probability_turn_intended))
+
+        self.grid.horz_wall(1, 7,  9, obj_type=SlipperyEast("white", probability_intended=self.probability_intended, probability_turn_intended=self.probability_turn_intended))
+        self.grid.horz_wall(1, 8,  8, obj_type=SlipperyEast("white", probability_intended=self.probability_intended, probability_turn_intended=self.probability_turn_intended))
+        self.grid.horz_wall(1, 9,  8, obj_type=SlipperyEast("white", probability_intended=self.probability_intended, probability_turn_intended=self.probability_turn_intended))
+        self.grid.horz_wall(1, 10, 7, obj_type=SlipperyEast("white", probability_intended=self.probability_intended, probability_turn_intended=self.probability_turn_intended))
+
+        self._place_building(16, 1, 4, 2)
+        self.grid.vert_wall(15, 1, 2, obj_type=SlipperyNorth("white", probability_intended=self.probability_intended, probability_turn_intended=self.probability_turn_intended))
+        self.grid.horz_wall(16, 3, 4, obj_type=SlipperyEast("white", probability_intended=self.probability_intended, probability_turn_intended=self.probability_turn_intended))
+        self.grid.vert_wall(20, 1, 2, obj_type=SlipperyNorth("white", probability_intended=self.probability_intended, probability_turn_intended=self.probability_turn_intended))
+
+        self._place_building(10,  3, 3, 4)
+        #self.grid.vert_wall(9, 3, 4, obj_type=SlipperyNorth("white", probability_intended=self.probability_intended, probability_turn_intended=self.probability_turn_intended))
+        self.grid.vert_wall(13, 3, 4, obj_type=SlipperyNorth("white", probability_intended=self.probability_intended, probability_turn_intended=self.probability_turn_intended))
+        self.grid.horz_wall(10, 2, 3, obj_type=SlipperyEast("white", probability_intended=self.probability_intended, probability_turn_intended=self.probability_turn_intended))
+        self.grid.horz_wall(10, 7, 3, obj_type=SlipperyEast("white", probability_intended=self.probability_intended, probability_turn_intended=self.probability_turn_intended))
+
+        self._place_building(16, 7, 8, 5)
+        self.grid.vert_wall(15, 7, 4, obj_type=SlipperyNorth("white", probability_intended=self.probability_intended, probability_turn_intended=self.probability_turn_intended))
+        #self.grid.vert_wall(17, 9, 3, obj_type=Lava)
+        self.grid.vert_wall(24, 7, 5, obj_type=SlipperyNorth("white", probability_intended=self.probability_intended, probability_turn_intended=self.probability_turn_intended))
+        self.grid.vert_wall(15, 9, 3, obj_type=SlipperyNorth("white", probability_intended=self.probability_intended, probability_turn_intended=self.probability_turn_intended))
+        self.grid.horz_wall(16, 6, 7, obj_type=SlipperyEast("white", probability_intended=self.probability_intended, probability_turn_intended=self.probability_turn_intended))
+        self.grid.horz_wall(16, 12, 7, obj_type=SlipperyEast("white", probability_intended=self.probability_intended, probability_turn_intended=self.probability_turn_intended))
+        self.grid.vert_wall(22, 12, 1, obj_type=SlipperyNorthEast("white", probability_intended=self.probability_intended, probability_turn_intended=self.probability_turn_intended))
+        self.grid.vert_wall(23, 13, 1, obj_type=SlipperyNorthEast("white", probability_intended=self.probability_intended, probability_turn_intended=self.probability_turn_intended))
+
+
+        self._place_building(8, 11, 5, 6)
+        #self.grid.vert_wall(7, 11, 6, obj_type=SlipperyNorth("white", probability_intended=self.probability_intended, probability_turn_intended=self.probability_turn_intended))
+        self.grid.vert_wall(13, 11, 6, obj_type=SlipperyNorth("white", probability_intended=self.probability_intended, probability_turn_intended=self.probability_turn_intended))
+        self.grid.horz_wall(8, 17, 5, obj_type=SlipperyEast("white", probability_intended=self.probability_intended, probability_turn_intended=self.probability_turn_intended))
+        self.grid.horz_wall(9, 10, 4, obj_type=SlipperyWest("white", probability_intended=self.probability_intended, probability_turn_intended=self.probability_turn_intended))
+        self.grid.horz_wall(10, 9, 3, obj_type=SlipperyWest("white", probability_intended=self.probability_intended, probability_turn_intended=self.probability_turn_intended))
+        self.grid.vert_wall(12, 7, 4, obj_type=SlipperySouth("white", probability_intended=self.probability_intended, probability_turn_intended=self.probability_turn_intended))
+
+        self._place_building(22, 14, 4, 4)
+        self.grid.vert_wall(21, 14, 4, obj_type=SlipperyNorth("white", probability_intended=self.probability_intended, probability_turn_intended=self.probability_turn_intended))
+        self.grid.horz_wall(22, 13, 4, obj_type=SlipperyEast("white", probability_intended=self.probability_intended, probability_turn_intended=self.probability_turn_intended))
+        self.grid.horz_wall(22, 18, 4, obj_type=SlipperyEast("white", probability_intended=self.probability_intended, probability_turn_intended=self.probability_turn_intended))
+
+
+        #self.grid.vert_wall(22, 13, 1, obj_type=SlipperyNorthEast("white", probability_intended=self.probability_intended, probability_turn_intended=self.probability_turn_intended))
+
+
+
+        self.place_agent(agent_pos=np.array((18, height - 4)), agent_dir=3, spawn_on_slippery=True)
+        self.place_goal(np.array((width - 2, 10)))
+        if self.dense_rewards: self.run_bfs()
+
+
+
+
+
+class WindyCitySmallAdv(AdversaryEnv):
+    def __init__(self,
+                randomize_start=True, size=10,
+                width=11,
+                height=9,
+                probability_intended=1,
+                probability_turn_intended=1,
+                obstacle_type=Lava,
+                goal_reward=1,
+                failure_penalty=-1,
+                per_step_penalty=0,
+                dense_rewards=False,
+                two_player_winning_region_start=False,
+                     **kwargs):
+
+        self.obstacle_type = obstacle_type
+        self.size = size
+        self.probability_intended = probability_intended
+        self.probability_turn_intended = probability_turn_intended
+
+        if width is not None and height is not None:
+            self.width = width
+            self.height = height
+        elif size is not None:
+            self.width = size
+            self.height = size
+        else:
+            raise ValueError(f"Please define either width and height or a size for square environments. The set values are width: {width}, height: {height}, size: {size}.")
+
+        mission_space = MissionSpace(mission_func=self._gen_mission)
+        super().__init__(
+            width=self.width,
+            height=self.height,
+            max_steps=50,
+            # Set this to True for maximum speed
+            see_through_walls=False,
+            #mission_space = mission_space,
+            **kwargs
+        )
+
+        self.randomize_start = randomize_start
+        self.two_player_winning_region_start = two_player_winning_region_start
+        self.goal_reward = goal_reward
+        self.failure_penalty = failure_penalty
+        self.dense_rewards = dense_rewards
+        self.per_step_penalty = per_step_penalty
+
+        self.trajectory = list()
+
+    @staticmethod
+    def _gen_mission():
+        return "Finish your task while avoiding the adversaries"
+
+    def disable_random_start(self):
+        self.randomize_start = False
+
+    def place_agent(self, spawn_on_slippery=False, agent_pos=None, agent_dir=0):
+        max_tries = 10_000
+        num_tries = 0
+
+        if self.randomize_start == True:
+            while True:
+                num_tries += 1
+                if num_tries > max_tries:
+                    raise RecursionError("rejection sampling failed in place_obj")
+                x = np.random.randint(0, self.width)
+                y = np.random.randint(5, self.height)
+
+                cell = self.grid.get(*(x,y))
+                if ( cell is None or
+                    (cell.can_overlap() and
+                        not isinstance(cell, Lava) and
+                        not isinstance(cell, Goal) and
+                        (spawn_on_slippery or not is_slippery(cell)) and
+                        not (x in [7, 8, 9, 10] and y in [9, 10]))
+                    ):
+                    self.agent_pos = np.array((x, y))
+                    self.agent_dir = np.random.randint(0, 4)
+                    break
+        elif agent_dir is None:
+            self.agent_pos = np.array((1, 1))
+            self.agent_dir = 0
+        else:
+            self.agent_pos = agent_pos
+            self.agent_dir = agent_dir
+        self.trajectory.append((self.agent_pos, self.agent_dir))
+
+    def place_goal(self, goal_pos):
+        self.goal_pos = goal_pos
+        self.put_obj(Goal(), *self.goal_pos)
+
+
+    def printGrid(self, init=False):
+        grid = super().printGrid(init)
+
+        properties_str = ""
+
+        properties_str += F"ProbTurnIntended:{self.probability_turn_intended}\n"
+        properties_str += F"ProbForwardIntended:{self.probability_intended}\n"
+
+        return grid + properties_str
+
+    def step(self, action):
+        obs, reward, terminated, truncated, info = super().step(action)
+        self.trajectory.append((action, self.agent_pos, self.agent_dir))
+        if truncated and info["ran_into_lava"]:
+            print(self.trajectory)
+            print("truncated: ", info)
+            self.trajectory = list()
+        if truncated and info["reached_goal"]:
+            print("truncated: ", info)
+            self.trajectory = list()
+        elif terminated and info["ran_into_lava"]:
+            print(self.trajectory)
+            print("terminated: ", info)
+            self.trajectory = list()
+        elif terminated and info["collision"]:
+            print(self.trajectory)
+            print("terminated: ", info)
+            self.trajectory = list()
+        elif terminated:
+            print("terminated: ", info)
+            self.trajectory = list()
+        elif truncated:
+            print("truncated: ", info)
+            self.trajectory = list()
+        return obs, reward - self.per_step_penalty, terminated, truncated, info
+
+    def reset(self, **kwargs) -> tuple[ObsType, dict[str, Any]]:
+        return super().reset(**kwargs)
+
+    def _place_building(self, col, row, width, height, obj_type=Lava):
+        for i in range(col, width + col):
+            self.grid.vert_wall(i, row, height, obj_type=obj_type)
+
+    def _gen_grid(self, width, height):
+        super()._gen_grid(width, height)
+        self.grid = Grid(width, height)
+
+        # Generate the surrounding walls
+        self.grid.horz_wall(0, 0)
+        self.grid.horz_wall(0, height - 1)
+        self.grid.vert_wall(0, 0)
+        self.grid.vert_wall(width - 1, 0)
+
+
+        self._place_building(3, 3, 5, 2)
+        blue_adv = self.add_adversary(2, 4, "blue", direction=3, tasks=
+                                      [GoTo((2,2)), GoTo((8,2)), GoTo((8,4)), GoTo((8,2)), GoTo((2,2)), GoTo((2,4))], repeating=True)
+
+
+        self.place_agent(agent_pos=np.array((5, 5)), agent_dir=3, spawn_on_slippery=True)
+        self.place_goal(np.array((width//2, 1)))
+        if self.dense_rewards: self.run_bfs()
+
+
+

notebooks/shielding_files_20241129T122425_0tlvpkoj/grid.txt

@@ -0,0 +1,49 @@
+WGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWG
+WGbWbWbWbWbWbWbWbWbWbWbWbWbWbWnWVRVRVRVRnWbWbWbWbWbWWG
+WGbWbWbWbWbWbWbWbWbWeWeWeWbWbWnWVRVRVRVRnWbWbWbWbWbWWG
+WGbWbWbWbWbWbWbWbWbWVRVRVRnWbWbWeWeWeWeWbWbWbWbWbWbWWG
+WGbWbWbWbWbWbWbWbWbWVRVRVRnWbWbWbWbWbWbWbWbWbWbWbWbWWG
+WGbWbWbWbWbWbWbWbWbWVRVRVRnWbWbWbWbWbWbWbWbWbWbWbWbWWG
+WGbWbWbWbWbWbWbWbWbWVRVRVRnWbWbWeWeWeWeWeWeWeWbWbWbWWG
+WGeWeWeWeWeWeWeWeWeWeWeWsWbWbWnWVRVRVRVRVRVRVRVRnWbWWG
+WGeWeWeWeWeWeWeWeWbWbWbWsWbWbWnWVRVRVRVRVRVRVRVRnWbWWG
+WGeWeWeWeWeWeWeWeWbWwWwWsWbWbWnWVRVRVRVRVRVRVRVRnWbWWG
+WGeWeWeWeWeWeWeWbWwWwWwWsWbWbWnWVRVRVRVRVRVRVRVRnWGGWG
+WGbWbWbWbWbWbWbWVRVRVRVRVRnWbWnWVRVRVRVRVRVRVRVRnWbWWG
+WGbWbWbWbWbWbWbWVRVRVRVRVRnWbWbWeWeWeWeWeWeWbWbWbWbWWG
+WGbWbWbWbWbWbWbWVRVRVRVRVRnWbWbWbWbWbWbWbWbWeWeWeWeWWG
+WGbWbWbWbWbWbWbWVRVRVRVRVRnWbWbWbWbWbWbWbWnWVRVRVRVRWG
+WGbWbWbWbWbWbWbWVRVRVRVRVRnWbWbWbWbWbWbWbWnWVRVRVRVRWG
+WGbWbWbWbWbWbWbWVRVRVRVRVRnWbWbWbWbWbWbWbWnWVRVRVRVRWG
+WGeWeWeWeWeWeWeWeWeWeWeWeWeWXReWbWbWbWbWbWnWVRVRVRVRWG
+WGeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWbWbWbWbWbWeWeWeWeWWG
+WGeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWbWbWbWbWbWbWbWbWWG
+WGeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWbWbWbWbWbWbWbWWG
+WGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWG
+------------------------------------------------------
+WGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWG
+WGbWbWbWbWbWbWbWbWbWbWbWbWbWbWnW        nWbWbWbWbWbWWG
+WGbWbWbWbWbWbWbWbWbWeWeWeWbWbWnW        nWbWbWbWbWbWWG
+WGbWbWbWbWbWbWbWbWbW      nWbWbWeWeWeWeWbWbWbWbWbWbWWG
+WGbWbWbWbWbWbWbWbWbW      nWbWbWbWbWbWbWbWbWbWbWbWbWWG
+WGbWbWbWbWbWbWbWbWbW      nWbWbWbWbWbWbWbWbWbWbWbWbWWG
+WGbWbWbWbWbWbWbWbWbW      nWbWbWeWeWeWeWeWeWeWbWbWbWWG
+WGeWeWeWeWeWeWeWeWeWeWeWsWbWbWnW                nWbWWG
+WGeWeWeWeWeWeWeWeWbWbWbWsWbWbWnW                nWbWWG
+WGeWeWeWeWeWeWeWeWbWwWwWsWbWbWnW                nWbWWG
+WGeWeWeWeWeWeWeWbWwWwWwWsWbWbWnW                nW  WG
+WGbWbWbWbWbWbWbW          nWbWnW                nWbWWG
+WGbWbWbWbWbWbWbW          nWbWbWeWeWeWeWeWeWbWbWbWbWWG
+WGbWbWbWbWbWbWbW          nWbWbWbWbWbWbWbWbWeWeWeWeWWG
+WGbWbWbWbWbWbWbW          nWbWbWbWbWbWbWbWnW        WG
+WGbWbWbWbWbWbWbW          nWbWbWbWbWbWbWbWnW        WG
+WGbWbWbWbWbWbWbW          nWbWbWbWbWbWbWbWnW        WG
+WGeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWbWbWbWbWbWnW        WG
+WGeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWbWbWbWbWbWeWeWeWeWWG
+WGeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWbWbWbWbWbWbWbWbWWG
+WGeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWbWbWbWbWbWbWbWWG
+WGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWG
+------------------------------------------------------
+------------------------------------------------------
+ProbTurnIntended:0.95
+ProbForwardIntended:0.95

notebooks/shielding_files_20241129T122427_qaabh7px/grid.txt

@@ -0,0 +1,49 @@
+WGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWG
+WGbWbWbWbWbWbWbWbWbWbWbWbWbWbWnWVRVRVRVRnWbWbWbWbWbWWG
+WGbWbWbWbWbWbWbWbWbWeWeWeWbWbWnWVRVRVRVRnWbWbWbWbWbWWG
+WGbWbWbWbWbWbWbWbWbWVRVRVRnWbWbWeWeWeWeWbWbWbWbWbWbWWG
+WGbWbWbWbWbWbWbWbWbWVRVRVRnWbWbWbWbWbWbWbWbWbWbWbWbWWG
+WGbWbWbWbWbWbWbWbWbWVRVRVRnWbWbWbWbWbWbWbWbWbWbWbWbWWG
+WGbWbWbWbWbWbWbWbWbWVRVRVRnWbWbWeWeWeWeWeWeWeWbWbWbWWG
+WGeWeWeWeWeWeWeWeWXReWeWsWbWbWnWVRVRVRVRVRVRVRVRnWbWWG
+WGeWeWeWeWeWeWeWeWbWbWbWsWbWbWnWVRVRVRVRVRVRVRVRnWbWWG
+WGeWeWeWeWeWeWeWeWbWwWwWsWbWbWnWVRVRVRVRVRVRVRVRnWbWWG
+WGeWeWeWeWeWeWeWbWwWwWwWsWbWbWnWVRVRVRVRVRVRVRVRnWGGWG
+WGbWbWbWbWbWbWbWVRVRVRVRVRnWbWnWVRVRVRVRVRVRVRVRnWbWWG
+WGbWbWbWbWbWbWbWVRVRVRVRVRnWbWbWeWeWeWeWeWeWbWbWbWbWWG
+WGbWbWbWbWbWbWbWVRVRVRVRVRnWbWbWbWbWbWbWbWbWeWeWeWeWWG
+WGbWbWbWbWbWbWbWVRVRVRVRVRnWbWbWbWbWbWbWbWnWVRVRVRVRWG
+WGbWbWbWbWbWbWbWVRVRVRVRVRnWbWbWbWbWbWbWbWnWVRVRVRVRWG
+WGbWbWbWbWbWbWbWVRVRVRVRVRnWbWbWbWbWbWbWbWnWVRVRVRVRWG
+WGeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWbWbWbWbWbWnWVRVRVRVRWG
+WGeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWbWbWbWbWbWeWeWeWeWWG
+WGeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWbWbWbWbWbWbWbWbWWG
+WGeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWbWbWbWbWbWbWbWWG
+WGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWG
+------------------------------------------------------
+WGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWG
+WGbWbWbWbWbWbWbWbWbWbWbWbWbWbWnW        nWbWbWbWbWbWWG
+WGbWbWbWbWbWbWbWbWbWeWeWeWbWbWnW        nWbWbWbWbWbWWG
+WGbWbWbWbWbWbWbWbWbW      nWbWbWeWeWeWeWbWbWbWbWbWbWWG
+WGbWbWbWbWbWbWbWbWbW      nWbWbWbWbWbWbWbWbWbWbWbWbWWG
+WGbWbWbWbWbWbWbWbWbW      nWbWbWbWbWbWbWbWbWbWbWbWbWWG
+WGbWbWbWbWbWbWbWbWbW      nWbWbWeWeWeWeWeWeWeWbWbWbWWG
+WGeWeWeWeWeWeWeWeWeWeWeWsWbWbWnW                nWbWWG
+WGeWeWeWeWeWeWeWeWbWbWbWsWbWbWnW                nWbWWG
+WGeWeWeWeWeWeWeWeWbWwWwWsWbWbWnW                nWbWWG
+WGeWeWeWeWeWeWeWbWwWwWwWsWbWbWnW                nW  WG
+WGbWbWbWbWbWbWbW          nWbWnW                nWbWWG
+WGbWbWbWbWbWbWbW          nWbWbWeWeWeWeWeWeWbWbWbWbWWG
+WGbWbWbWbWbWbWbW          nWbWbWbWbWbWbWbWbWeWeWeWeWWG
+WGbWbWbWbWbWbWbW          nWbWbWbWbWbWbWbWnW        WG
+WGbWbWbWbWbWbWbW          nWbWbWbWbWbWbWbWnW        WG
+WGbWbWbWbWbWbWbW          nWbWbWbWbWbWbWbWnW        WG
+WGeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWbWbWbWbWbWnW        WG
+WGeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWbWbWbWbWbWeWeWeWeWWG
+WGeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWbWbWbWbWbWbWbWbWWG
+WGeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWbWbWbWbWbWbWbWWG
+WGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWG
+------------------------------------------------------
+------------------------------------------------------
+ProbTurnIntended:0.95
+ProbForwardIntended:0.95

notebooks/shielding_files_20241129T122429_qyepifjg/grid.txt

@@ -0,0 +1,49 @@
+WGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWG
+WGbWbWbWbWbWbWbWbWbWbWbWbWbWbWnWVRVRVRVRnWbWbWbWbWbWWG
+WGbWbWbWbWbWbWbWbWbWeWeWeWbWbWnWVRVRVRVRnWbWbWbWbWbWWG
+WGbWbWbWbWbWbWbWbWbWVRVRVRnWbWbWeWeWeWeWbWbWbWbWbWbWWG
+WGbWbWbWbWbWbWbWbWbWVRVRVRnWbWbWbWbWbWbWbWbWbWbWbWbWWG
+WGbWbWbWbWbWbWbWbWbWVRVRVRnWbWbWbWbWbWbWbWbWbWbWbWbWWG
+WGbWbWbWbWbWbWbWbWbWVRVRVRnWbWbWeWeWeWeWeWeWeWbWbWbWWG
+WGeWeWeWeWeWeWeWeWeWeWeWsWbWbWnWVRVRVRVRVRVRVRVRnWbWWG
+WGeWeWeWeWeWeWeWeWbWbWbWsWbWbWnWVRVRVRVRVRVRVRVRnWbWWG
+WGeWeWeWeWeWeWeWeWbWwWwWsWbWbWnWVRVRVRVRVRVRVRVRnWbWWG
+WGeWeWeWeWeWeWeWbWwWwWwWsWbWbWnWVRVRVRVRVRVRVRVRnWGGWG
+WGbWbWbWbWbWbWbWVRVRVRVRVRnWbWnWVRVRVRVRVRVRVRVRnWbWWG
+WGbWbWbWbWbWbWbWVRVRVRVRVRnWbWbWeWeWeWeWeWeWbWbWbWbWWG
+WGbWbWbWbWbWbWbWVRVRVRVRVRnWbWbWbWbWbWbWbWbWeWeWeWeWWG
+WGbWbWbWbWbWbWbWVRVRVRVRVRnWbWbWbWbWbWbWbWnWVRVRVRVRWG
+WGbWbWbWbWbWbWbWVRVRVRVRVRnWbWbWbWbWbWbWbWnWVRVRVRVRWG
+WGbWbWbWbWbWbWbWVRVRVRVRVRnWbWbWbWbWbWbWbWnWVRVRVRVRWG
+WGeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWbWbWbWbWbWnWVRVRVRVRWG
+WGeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWbWbWbWbWbWeWeWeWeWWG
+WGeWeWeWeWeWeWeWeWeWeWXReWeWeWeWeWeWbWbWbWbWbWbWbWbWWG
+WGeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWbWbWbWbWbWbWbWWG
+WGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWG
+------------------------------------------------------
+WGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWG
+WGbWbWbWbWbWbWbWbWbWbWbWbWbWbWnW        nWbWbWbWbWbWWG
+WGbWbWbWbWbWbWbWbWbWeWeWeWbWbWnW        nWbWbWbWbWbWWG
+WGbWbWbWbWbWbWbWbWbW      nWbWbWeWeWeWeWbWbWbWbWbWbWWG
+WGbWbWbWbWbWbWbWbWbW      nWbWbWbWbWbWbWbWbWbWbWbWbWWG
+WGbWbWbWbWbWbWbWbWbW      nWbWbWbWbWbWbWbWbWbWbWbWbWWG
+WGbWbWbWbWbWbWbWbWbW      nWbWbWeWeWeWeWeWeWeWbWbWbWWG
+WGeWeWeWeWeWeWeWeWeWeWeWsWbWbWnW                nWbWWG
+WGeWeWeWeWeWeWeWeWbWbWbWsWbWbWnW                nWbWWG
+WGeWeWeWeWeWeWeWeWbWwWwWsWbWbWnW                nWbWWG
+WGeWeWeWeWeWeWeWbWwWwWwWsWbWbWnW                nW  WG
+WGbWbWbWbWbWbWbW          nWbWnW                nWbWWG
+WGbWbWbWbWbWbWbW          nWbWbWeWeWeWeWeWeWbWbWbWbWWG
+WGbWbWbWbWbWbWbW          nWbWbWbWbWbWbWbWbWeWeWeWeWWG
+WGbWbWbWbWbWbWbW          nWbWbWbWbWbWbWbWnW        WG
+WGbWbWbWbWbWbWbW          nWbWbWbWbWbWbWbWnW        WG
+WGbWbWbWbWbWbWbW          nWbWbWbWbWbWbWbWnW        WG
+WGeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWbWbWbWbWbWnW        WG
+WGeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWbWbWbWbWbWeWeWeWeWWG
+WGeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWbWbWbWbWbWbWbWbWWG
+WGeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWbWbWbWbWbWbWbWWG
+WGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWG
+------------------------------------------------------
+------------------------------------------------------
+ProbTurnIntended:0.95
+ProbForwardIntended:0.95

notebooks/shielding_files_20241129T122432_3ui3yv2s/grid.txt

@@ -0,0 +1,49 @@
+WGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWG
+WGbWbWbWbWbWbWbWbWbWbWbWbWbWbWnWVRVRVRVRnWbWbWbWbWbWWG
+WGbWbWbWbWbWbWbWbWbWeWeWeWbWbWnWVRVRVRVRnWbWbWbWbWbWWG
+WGbWbWbWbWbWbWbWbWbWVRVRVRnWbWbWeWeWeWeWbWbWbWbWbWbWWG
+WGbWbWbWbWbWbWbWbWbWVRVRVRnWbWbWbWbWbWbWbWbWbWbWbWbWWG
+WGbWbWbWbWbWbWbWbWbWVRVRVRnWbWbWbWbWbWbWbWbWbWbWbWbWWG
+WGbWbWbWbWbWbWbWbWbWVRVRVRnWbWbWeWeWeWeWeWeWeWbWbWbWWG
+WGeWeWeWeWeWeWeWeWeWeWeWsWbWbWnWVRVRVRVRVRVRVRVRnWbWWG
+WGeWeWeWeWeWeWeWeWbWbWbWsWbWbWnWVRVRVRVRVRVRVRVRnWbWWG
+WGeWeWeWeWeWeWeWeWbWwWwWsWbWbWnWVRVRVRVRVRVRVRVRnWbWWG
+WGeWeWeWeWeWeWeWbWwWwWwWsWbWbWnWVRVRVRVRVRVRVRVRnWGGWG
+WGbWbWbWbWbWbWbWVRVRVRVRVRnWbWnWVRVRVRVRVRVRVRVRnWbWWG
+WGbWbWbWbWbWbWbWVRVRVRVRVRnWbWbWeWeWeWeWeWeWbWbWbWbWWG
+WGbWbWbWbWbWbWbWVRVRVRVRVRnWbWbWbWbWbWbWbWbWeWeWeWeWWG
+WGbWbWbWbWbWbWbWVRVRVRVRVRnWbWbWbWbWbWbWXRnWVRVRVRVRWG
+WGbWbWbWbWbWbWbWVRVRVRVRVRnWbWbWbWbWbWbWbWnWVRVRVRVRWG
+WGbWbWbWbWbWbWbWVRVRVRVRVRnWbWbWbWbWbWbWbWnWVRVRVRVRWG
+WGeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWbWbWbWbWbWnWVRVRVRVRWG
+WGeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWbWbWbWbWbWeWeWeWeWWG
+WGeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWbWbWbWbWbWbWbWbWWG
+WGeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWbWbWbWbWbWbWbWWG
+WGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWG
+------------------------------------------------------
+WGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWG
+WGbWbWbWbWbWbWbWbWbWbWbWbWbWbWnW        nWbWbWbWbWbWWG
+WGbWbWbWbWbWbWbWbWbWeWeWeWbWbWnW        nWbWbWbWbWbWWG
+WGbWbWbWbWbWbWbWbWbW      nWbWbWeWeWeWeWbWbWbWbWbWbWWG
+WGbWbWbWbWbWbWbWbWbW      nWbWbWbWbWbWbWbWbWbWbWbWbWWG
+WGbWbWbWbWbWbWbWbWbW      nWbWbWbWbWbWbWbWbWbWbWbWbWWG
+WGbWbWbWbWbWbWbWbWbW      nWbWbWeWeWeWeWeWeWeWbWbWbWWG
+WGeWeWeWeWeWeWeWeWeWeWeWsWbWbWnW                nWbWWG
+WGeWeWeWeWeWeWeWeWbWbWbWsWbWbWnW                nWbWWG
+WGeWeWeWeWeWeWeWeWbWwWwWsWbWbWnW                nWbWWG
+WGeWeWeWeWeWeWeWbWwWwWwWsWbWbWnW                nW  WG
+WGbWbWbWbWbWbWbW          nWbWnW                nWbWWG
+WGbWbWbWbWbWbWbW          nWbWbWeWeWeWeWeWeWbWbWbWbWWG
+WGbWbWbWbWbWbWbW          nWbWbWbWbWbWbWbWbWeWeWeWeWWG
+WGbWbWbWbWbWbWbW          nWbWbWbWbWbWbWbWnW        WG
+WGbWbWbWbWbWbWbW          nWbWbWbWbWbWbWbWnW        WG
+WGbWbWbWbWbWbWbW          nWbWbWbWbWbWbWbWnW        WG
+WGeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWbWbWbWbWbWnW        WG
+WGeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWbWbWbWbWbWeWeWeWeWWG
+WGeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWbWbWbWbWbWbWbWbWWG
+WGeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWbWbWbWbWbWbWbWWG
+WGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWG
+------------------------------------------------------
+------------------------------------------------------
+ProbTurnIntended:0.95
+ProbForwardIntended:0.95

notebooks/shielding_files_20241129T122454_58hszne6/grid.txt

@@ -0,0 +1,49 @@
+WGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWG
+WGbWbWbWbWbWbWbWbWbWbWbWbWbWbWnWVRVRVRVRnWbWbWbWbWbWWG
+WGbWbWbWbWbWbWbWbWbWeWeWeWbWbWnWVRVRVRVRnWbWbWbWbWbWWG
+WGbWbWbWbWbWbWbWbWbWVRVRVRnWbWbWeWeWeWeWbWbWbWbWbWbWWG
+WGbWbWbWbWbWbWbWbWbWVRVRVRnWbWbWbWbWbWbWbWbWbWbWbWbWWG
+WGbWbWbWbWbWbWbWbWbWVRVRVRnWbWbWbWbWbWbWbWbWbWbWbWbWWG
+WGbWbWbWbWbWbWbWbWbWVRVRVRnWbWbWeWeWeWeWeWeWeWbWbWbWWG
+WGXReWeWeWeWeWeWeWeWeWeWsWbWbWnWVRVRVRVRVRVRVRVRnWbWWG
+WGeWeWeWeWeWeWeWeWbWbWbWsWbWbWnWVRVRVRVRVRVRVRVRnWbWWG
+WGeWeWeWeWeWeWeWeWbWwWwWsWbWbWnWVRVRVRVRVRVRVRVRnWbWWG
+WGeWeWeWeWeWeWeWbWwWwWwWsWbWbWnWVRVRVRVRVRVRVRVRnWGGWG
+WGbWbWbWbWbWbWbWVRVRVRVRVRnWbWnWVRVRVRVRVRVRVRVRnWbWWG
+WGbWbWbWbWbWbWbWVRVRVRVRVRnWbWbWeWeWeWeWeWeWbWbWbWbWWG
+WGbWbWbWbWbWbWbWVRVRVRVRVRnWbWbWbWbWbWbWbWbWeWeWeWeWWG
+WGbWbWbWbWbWbWbWVRVRVRVRVRnWbWbWbWbWbWbWbWnWVRVRVRVRWG
+WGbWbWbWbWbWbWbWVRVRVRVRVRnWbWbWbWbWbWbWbWnWVRVRVRVRWG
+WGbWbWbWbWbWbWbWVRVRVRVRVRnWbWbWbWbWbWbWbWnWVRVRVRVRWG
+WGeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWbWbWbWbWbWnWVRVRVRVRWG
+WGeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWbWbWbWbWbWeWeWeWeWWG
+WGeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWbWbWbWbWbWbWbWbWWG
+WGeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWbWbWbWbWbWbWbWWG
+WGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWG
+------------------------------------------------------
+WGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWG
+WGbWbWbWbWbWbWbWbWbWbWbWbWbWbWnW        nWbWbWbWbWbWWG
+WGbWbWbWbWbWbWbWbWbWeWeWeWbWbWnW        nWbWbWbWbWbWWG
+WGbWbWbWbWbWbWbWbWbW      nWbWbWeWeWeWeWbWbWbWbWbWbWWG
+WGbWbWbWbWbWbWbWbWbW      nWbWbWbWbWbWbWbWbWbWbWbWbWWG
+WGbWbWbWbWbWbWbWbWbW      nWbWbWbWbWbWbWbWbWbWbWbWbWWG
+WGbWbWbWbWbWbWbWbWbW      nWbWbWeWeWeWeWeWeWeWbWbWbWWG
+WGeWeWeWeWeWeWeWeWeWeWeWsWbWbWnW                nWbWWG
+WGeWeWeWeWeWeWeWeWbWbWbWsWbWbWnW                nWbWWG
+WGeWeWeWeWeWeWeWeWbWwWwWsWbWbWnW                nWbWWG
+WGeWeWeWeWeWeWeWbWwWwWwWsWbWbWnW                nW  WG
+WGbWbWbWbWbWbWbW          nWbWnW                nWbWWG
+WGbWbWbWbWbWbWbW          nWbWbWeWeWeWeWeWeWbWbWbWbWWG
+WGbWbWbWbWbWbWbW          nWbWbWbWbWbWbWbWbWeWeWeWeWWG
+WGbWbWbWbWbWbWbW          nWbWbWbWbWbWbWbWnW        WG
+WGbWbWbWbWbWbWbW          nWbWbWbWbWbWbWbWnW        WG
+WGbWbWbWbWbWbWbW          nWbWbWbWbWbWbWbWnW        WG
+WGeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWbWbWbWbWbWnW        WG
+WGeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWbWbWbWbWbWeWeWeWeWWG
+WGeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWbWbWbWbWbWbWbWbWWG
+WGeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWbWbWbWbWbWbWbWWG
+WGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWG
+------------------------------------------------------
+------------------------------------------------------
+ProbTurnIntended:0.95
+ProbForwardIntended:0.95

notebooks/shielding_files_20241129T122456_wtrsvp2x/grid.txt

@@ -0,0 +1,49 @@
+WGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWG
+WGbWbWbWbWbWbWbWbWbWbWbWbWbWbWnWVRVRVRVRnWbWbWbWbWbWWG
+WGbWbWbWbWbWbWbWbWbWeWeWeWbWbWnWVRVRVRVRnWbWbWbWbWbWWG
+WGbWbWbWbWbWbWbWbWbWVRVRVRnWbWbWeWeWeWeWbWbWbWbWbWbWWG
+WGbWbWbWbWbWbWbWbWbWVRVRVRnWbWbWbWbWbWbWbWbWbWbWbWbWWG
+WGbWbWbWbWbWbWbWbWbWVRVRVRnWbWbWbWbWbWbWbWbWbWbWbWbWWG
+WGbWbWbWbWbWbWbWXRbWVRVRVRnWbWbWeWeWeWeWeWeWeWbWbWbWWG
+WGeWeWeWeWeWeWeWeWeWeWeWsWbWbWnWVRVRVRVRVRVRVRVRnWbWWG
+WGeWeWeWeWeWeWeWeWbWbWbWsWbWbWnWVRVRVRVRVRVRVRVRnWbWWG
+WGeWeWeWeWeWeWeWeWbWwWwWsWbWbWnWVRVRVRVRVRVRVRVRnWbWWG
+WGeWeWeWeWeWeWeWbWwWwWwWsWbWbWnWVRVRVRVRVRVRVRVRnWGGWG
+WGbWbWbWbWbWbWbWVRVRVRVRVRnWbWnWVRVRVRVRVRVRVRVRnWbWWG
+WGbWbWbWbWbWbWbWVRVRVRVRVRnWbWbWeWeWeWeWeWeWbWbWbWbWWG
+WGbWbWbWbWbWbWbWVRVRVRVRVRnWbWbWbWbWbWbWbWbWeWeWeWeWWG
+WGbWbWbWbWbWbWbWVRVRVRVRVRnWbWbWbWbWbWbWbWnWVRVRVRVRWG
+WGbWbWbWbWbWbWbWVRVRVRVRVRnWbWbWbWbWbWbWbWnWVRVRVRVRWG
+WGbWbWbWbWbWbWbWVRVRVRVRVRnWbWbWbWbWbWbWbWnWVRVRVRVRWG
+WGeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWbWbWbWbWbWnWVRVRVRVRWG
+WGeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWbWbWbWbWbWeWeWeWeWWG
+WGeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWbWbWbWbWbWbWbWbWWG
+WGeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWbWbWbWbWbWbWbWWG
+WGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWG
+------------------------------------------------------
+WGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWG
+WGbWbWbWbWbWbWbWbWbWbWbWbWbWbWnW        nWbWbWbWbWbWWG
+WGbWbWbWbWbWbWbWbWbWeWeWeWbWbWnW        nWbWbWbWbWbWWG
+WGbWbWbWbWbWbWbWbWbW      nWbWbWeWeWeWeWbWbWbWbWbWbWWG
+WGbWbWbWbWbWbWbWbWbW      nWbWbWbWbWbWbWbWbWbWbWbWbWWG
+WGbWbWbWbWbWbWbWbWbW      nWbWbWbWbWbWbWbWbWbWbWbWbWWG
+WGbWbWbWbWbWbWbWbWbW      nWbWbWeWeWeWeWeWeWeWbWbWbWWG
+WGeWeWeWeWeWeWeWeWeWeWeWsWbWbWnW                nWbWWG
+WGeWeWeWeWeWeWeWeWbWbWbWsWbWbWnW                nWbWWG
+WGeWeWeWeWeWeWeWeWbWwWwWsWbWbWnW                nWbWWG
+WGeWeWeWeWeWeWeWbWwWwWwWsWbWbWnW                nW  WG
+WGbWbWbWbWbWbWbW          nWbWnW                nWbWWG
+WGbWbWbWbWbWbWbW          nWbWbWeWeWeWeWeWeWbWbWbWbWWG
+WGbWbWbWbWbWbWbW          nWbWbWbWbWbWbWbWbWeWeWeWeWWG
+WGbWbWbWbWbWbWbW          nWbWbWbWbWbWbWbWnW        WG
+WGbWbWbWbWbWbWbW          nWbWbWbWbWbWbWbWnW        WG
+WGbWbWbWbWbWbWbW          nWbWbWbWbWbWbWbWnW        WG
+WGeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWbWbWbWbWbWnW        WG
+WGeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWbWbWbWbWbWeWeWeWeWWG
+WGeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWbWbWbWbWbWbWbWbWWG
+WGeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWbWbWbWbWbWbWbWWG
+WGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWG
+------------------------------------------------------
+------------------------------------------------------
+ProbTurnIntended:0.95
+ProbForwardIntended:0.95

notebooks/shielding_files_20241129T122458_wavsk47p/grid.txt

@@ -0,0 +1,49 @@
+WGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWG
+WGbWbWbWbWbWbWbWbWbWbWbWbWbWbWnWVRVRVRVRnWbWbWbWbWbWWG
+WGbWbWbWbWbWbWbWbWbWeWeWeWbWbWnWVRVRVRVRnWbWbWXRbWbWWG
+WGbWbWbWbWbWbWbWbWbWVRVRVRnWbWbWeWeWeWeWbWbWbWbWbWbWWG
+WGbWbWbWbWbWbWbWbWbWVRVRVRnWbWbWbWbWbWbWbWbWbWbWbWbWWG
+WGbWbWbWbWbWbWbWbWbWVRVRVRnWbWbWbWbWbWbWbWbWbWbWbWbWWG
+WGbWbWbWbWbWbWbWbWbWVRVRVRnWbWbWeWeWeWeWeWeWeWbWbWbWWG
+WGeWeWeWeWeWeWeWeWeWeWeWsWbWbWnWVRVRVRVRVRVRVRVRnWbWWG
+WGeWeWeWeWeWeWeWeWbWbWbWsWbWbWnWVRVRVRVRVRVRVRVRnWbWWG
+WGeWeWeWeWeWeWeWeWbWwWwWsWbWbWnWVRVRVRVRVRVRVRVRnWbWWG
+WGeWeWeWeWeWeWeWbWwWwWwWsWbWbWnWVRVRVRVRVRVRVRVRnWGGWG
+WGbWbWbWbWbWbWbWVRVRVRVRVRnWbWnWVRVRVRVRVRVRVRVRnWbWWG
+WGbWbWbWbWbWbWbWVRVRVRVRVRnWbWbWeWeWeWeWeWeWbWbWbWbWWG
+WGbWbWbWbWbWbWbWVRVRVRVRVRnWbWbWbWbWbWbWbWbWeWeWeWeWWG
+WGbWbWbWbWbWbWbWVRVRVRVRVRnWbWbWbWbWbWbWbWnWVRVRVRVRWG
+WGbWbWbWbWbWbWbWVRVRVRVRVRnWbWbWbWbWbWbWbWnWVRVRVRVRWG
+WGbWbWbWbWbWbWbWVRVRVRVRVRnWbWbWbWbWbWbWbWnWVRVRVRVRWG
+WGeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWbWbWbWbWbWnWVRVRVRVRWG
+WGeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWbWbWbWbWbWeWeWeWeWWG
+WGeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWbWbWbWbWbWbWbWbWWG
+WGeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWbWbWbWbWbWbWbWWG
+WGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWG
+------------------------------------------------------
+WGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWG
+WGbWbWbWbWbWbWbWbWbWbWbWbWbWbWnW        nWbWbWbWbWbWWG
+WGbWbWbWbWbWbWbWbWbWeWeWeWbWbWnW        nWbWbWbWbWbWWG
+WGbWbWbWbWbWbWbWbWbW      nWbWbWeWeWeWeWbWbWbWbWbWbWWG
+WGbWbWbWbWbWbWbWbWbW      nWbWbWbWbWbWbWbWbWbWbWbWbWWG
+WGbWbWbWbWbWbWbWbWbW      nWbWbWbWbWbWbWbWbWbWbWbWbWWG
+WGbWbWbWbWbWbWbWbWbW      nWbWbWeWeWeWeWeWeWeWbWbWbWWG
+WGeWeWeWeWeWeWeWeWeWeWeWsWbWbWnW                nWbWWG
+WGeWeWeWeWeWeWeWeWbWbWbWsWbWbWnW                nWbWWG
+WGeWeWeWeWeWeWeWeWbWwWwWsWbWbWnW                nWbWWG
+WGeWeWeWeWeWeWeWbWwWwWwWsWbWbWnW                nW  WG
+WGbWbWbWbWbWbWbW          nWbWnW                nWbWWG
+WGbWbWbWbWbWbWbW          nWbWbWeWeWeWeWeWeWbWbWbWbWWG
+WGbWbWbWbWbWbWbW          nWbWbWbWbWbWbWbWbWeWeWeWeWWG
+WGbWbWbWbWbWbWbW          nWbWbWbWbWbWbWbWnW        WG
+WGbWbWbWbWbWbWbW          nWbWbWbWbWbWbWbWnW        WG
+WGbWbWbWbWbWbWbW          nWbWbWbWbWbWbWbWnW        WG
+WGeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWbWbWbWbWbWnW        WG
+WGeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWbWbWbWbWbWeWeWeWeWWG
+WGeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWbWbWbWbWbWbWbWbWWG
+WGeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWbWbWbWbWbWbWbWWG
+WGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWG
+------------------------------------------------------
+------------------------------------------------------
+ProbTurnIntended:0.95
+ProbForwardIntended:0.95

notebooks/shielding_files_20241129T122500_n10bceo8/grid.txt

@@ -0,0 +1,49 @@
+WGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWG
+WGbWbWbWbWbWbWbWbWbWbWbWbWbWbWnWVRVRVRVRnWbWbWbWbWbWWG
+WGbWbWbWbWbWbWbWbWbWeWeWeWbWbWnWVRVRVRVRnWbWbWbWbWbWWG
+WGbWbWbWbWbWbWbWbWbWVRVRVRnWbWbWeWeWeWeWbWbWbWbWbWbWWG
+WGXRbWbWbWbWbWbWbWbWVRVRVRnWbWbWbWbWbWbWbWbWbWbWbWbWWG
+WGbWbWbWbWbWbWbWbWbWVRVRVRnWbWbWbWbWbWbWbWbWbWbWbWbWWG
+WGbWbWbWbWbWbWbWbWbWVRVRVRnWbWbWeWeWeWeWeWeWeWbWbWbWWG
+WGeWeWeWeWeWeWeWeWeWeWeWsWbWbWnWVRVRVRVRVRVRVRVRnWbWWG
+WGeWeWeWeWeWeWeWeWbWbWbWsWbWbWnWVRVRVRVRVRVRVRVRnWbWWG
+WGeWeWeWeWeWeWeWeWbWwWwWsWbWbWnWVRVRVRVRVRVRVRVRnWbWWG
+WGeWeWeWeWeWeWeWbWwWwWwWsWbWbWnWVRVRVRVRVRVRVRVRnWGGWG
+WGbWbWbWbWbWbWbWVRVRVRVRVRnWbWnWVRVRVRVRVRVRVRVRnWbWWG
+WGbWbWbWbWbWbWbWVRVRVRVRVRnWbWbWeWeWeWeWeWeWbWbWbWbWWG
+WGbWbWbWbWbWbWbWVRVRVRVRVRnWbWbWbWbWbWbWbWbWeWeWeWeWWG
+WGbWbWbWbWbWbWbWVRVRVRVRVRnWbWbWbWbWbWbWbWnWVRVRVRVRWG
+WGbWbWbWbWbWbWbWVRVRVRVRVRnWbWbWbWbWbWbWbWnWVRVRVRVRWG
+WGbWbWbWbWbWbWbWVRVRVRVRVRnWbWbWbWbWbWbWbWnWVRVRVRVRWG
+WGeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWbWbWbWbWbWnWVRVRVRVRWG
+WGeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWbWbWbWbWbWeWeWeWeWWG
+WGeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWbWbWbWbWbWbWbWbWWG
+WGeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWbWbWbWbWbWbWbWWG
+WGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWG
+------------------------------------------------------
+WGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWG
+WGbWbWbWbWbWbWbWbWbWbWbWbWbWbWnW        nWbWbWbWbWbWWG
+WGbWbWbWbWbWbWbWbWbWeWeWeWbWbWnW        nWbWbWbWbWbWWG
+WGbWbWbWbWbWbWbWbWbW      nWbWbWeWeWeWeWbWbWbWbWbWbWWG
+WGbWbWbWbWbWbWbWbWbW      nWbWbWbWbWbWbWbWbWbWbWbWbWWG
+WGbWbWbWbWbWbWbWbWbW      nWbWbWbWbWbWbWbWbWbWbWbWbWWG
+WGbWbWbWbWbWbWbWbWbW      nWbWbWeWeWeWeWeWeWeWbWbWbWWG
+WGeWeWeWeWeWeWeWeWeWeWeWsWbWbWnW                nWbWWG
+WGeWeWeWeWeWeWeWeWbWbWbWsWbWbWnW                nWbWWG
+WGeWeWeWeWeWeWeWeWbWwWwWsWbWbWnW                nWbWWG
+WGeWeWeWeWeWeWeWbWwWwWwWsWbWbWnW                nW  WG
+WGbWbWbWbWbWbWbW          nWbWnW                nWbWWG
+WGbWbWbWbWbWbWbW          nWbWbWeWeWeWeWeWeWbWbWbWbWWG
+WGbWbWbWbWbWbWbW          nWbWbWbWbWbWbWbWbWeWeWeWeWWG
+WGbWbWbWbWbWbWbW          nWbWbWbWbWbWbWbWnW        WG
+WGbWbWbWbWbWbWbW          nWbWbWbWbWbWbWbWnW        WG
+WGbWbWbWbWbWbWbW          nWbWbWbWbWbWbWbWnW        WG
+WGeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWbWbWbWbWbWnW        WG
+WGeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWbWbWbWbWbWeWeWeWeWWG
+WGeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWbWbWbWbWbWbWbWbWWG
+WGeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWeWbWbWbWbWbWbWbWWG
+WGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWGWG
+------------------------------------------------------
+------------------------------------------------------
+ProbTurnIntended:0.95
+ProbForwardIntended:0.95