diff --git a/hw3/main.py b/hw3/main.py
new file mode 100644
index 00000000..921fc889
--- /dev/null
+++ b/hw3/main.py
@@ -0,0 +1,96 @@
+import random
+import sys
+from typing import List, Tuple
+
+from map_processor import load_and_filter_map, select_start, get_goal_pixels
+from navigator import init_sim, execute_waypoint_path
+
+
+POINT_CLOUD_DATA = "semantic_3d_pointcloud/point.npy"
+COLOR_DATA = "semantic_3d_pointcloud/color0255.npy"
+
+# Sample semantic color and index dictionaries for a few object categories. 
+# Check hw0/replica_v1/apartment_0/habitat/info_semantic.json and 
+# hw3/color_coding_semantic_segmentation_classes.xlsx for the full list of 
+# categories and their corresponding colors and indices.
+SEMANTIC_DICTS = {
+    "colors": {
+        "rack": [[0, 255, 133]],
+        "cushion": [[255, 9, 92]],
+        "sofa": [[10, 0, 255]],
+        "stair": [[173, 255, 0]],
+        "cooktop": [[7, 255, 224]],
+    },
+    "indices": {
+        "rack": 66,
+        "cushion": 29,
+        "sofa": 76,
+        "stair": 77,
+        "cooktop": 32,
+    },
+}
+
+
+def pick_goal(map_img) -> Tuple[str, Tuple[int, int]]:
+    prompt = "Enter semantic destination (ex: 'rack', 'cooktop', 'sofa'): "
+    goal_prompt = input(prompt).strip().lower()
+    if goal_prompt not in SEMANTIC_DICTS["colors"]:
+        print(f"Goal '{goal_prompt}' is not valid.")
+        sys.exit(1)
+
+    goal_pixels = get_goal_pixels(map_img, SEMANTIC_DICTS["colors"], goal_prompt)
+    goal = random.choice(goal_pixels)
+    return goal_prompt, goal
+
+
+def run_in_sim(start_world: Tuple[float, float], world_path: List[Tuple[float, float]], goal_prompt: str):
+    start_x, start_z = start_world
+    print(f"Spawning Agent at world position: ({start_x:.3f}, {start_z:.3f})")
+
+    sim, agent, _ = init_sim(start_x=start_x, start_z=start_z)
+    execute_waypoint_path(world_path, sim, agent, SEMANTIC_DICTS["indices"][goal_prompt])
+
+
+def main():
+    """Entry point."""
+
+    print("=== Step 1: Processing the 3D Map ===")
+    # =============== TODO 1-2 ===============
+    # map_img, occupancy_map, ... = load_and_filter_map(POINT_CLOUD_DATA, COLOR_DATA)
+
+
+    print("=== Step 2: Selecting Agent Start and Goal Positions ===")
+    start = select_start(map_img)
+    goal_prompt, goal = pick_goal(map_img)
+    print(f"Goal pixel selected at coordinates: {goal}")
+
+
+    print("=== Step 3: Executing Path Planning (RRT) ===")
+    # =============== TODO 2 ===============
+    # implement RRT path planner in plan_path()
+
+    # path = plan_path(start, goal, occupancy_map)
+    # if not path:
+    #     print("Planner could not find a path.")
+    #     sys.exit(1)
+
+
+    print("=== Step 4: Visualizing the Planned Path ===")
+    # =============== TODO 3 ===============
+    # Visualize the planned path over the map
+
+    # visualize_path(...)
+
+
+    print("=== Step 5: Translating Path to Habitat Simulator ===")
+    # =============== TODO 4 ===============
+    # Convert pixel path to world coordinates
+    # world_path is a list of tuples(float, float) representing waypoints in world coordinates
+
+    # world_path = ... 
+
+    run_in_sim(world_path[0], world_path, goal_prompt)
+
+
+if __name__ == "__main__":
+    main()