# Pothole Cluster Detection Simulation with DBSCAN
# Author: Yassir-style prototype

import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
from sklearn.cluster import DBSCAN
from sklearn.preprocessing import StandardScaler

# -----------------------------
# 1️⃣ Generate simulated vehicle data
# -----------------------------
def generate_vehicle_data(n_points=2000, n_potholes=4, city_center=(42.3314, -83.0458),
                          lat_spread=0.05, lon_spread=0.08, random_state=42):
    np.random.seed(random_state)
    
    # Base street distribution
    latitudes = city_center[0] + np.random.uniform(-lat_spread, lat_spread, n_points)
    longitudes = city_center[1] + np.random.uniform(-lon_spread, lon_spread, n_points)
    
    # Normal road conditions
    iri = np.random.normal(1.5, 0.3, n_points)          # International Roughness Index
    wheel_slip = np.random.normal(0.02, 0.01, n_points) # Normal wheel slip
    speed = np.random.normal(35, 5, n_points)           # Vehicle speed (mph)
    
    # Inject potholes
    pothole_centers = []
    for i in range(n_potholes):
        center_lat = city_center[0] + np.random.uniform(-lat_spread, lat_spread)
        center_lon = city_center[1] + np.random.uniform(-lon_spread, lon_spread)
        pothole_centers.append((center_lat, center_lon))
        
        # Assign random points to this pothole
        idx = np.random.choice(range(n_points), size=int(n_points*0.05), replace=False)
        latitudes[idx] = center_lat + np.random.normal(0, 0.001, len(idx))
        longitudes[idx] = center_lon + np.random.normal(0, 0.001, len(idx))
        iri[idx] = np.random.normal(4.5, 0.7, len(idx))
        wheel_slip[idx] = np.random.normal(0.12, 0.03, len(idx))
        speed[idx] = np.random.normal(18, 3, len(idx))
    
    df = pd.DataFrame({
        "latitude": latitudes,
        "longitude": longitudes,
        "iri": iri,
        "wheel_slip": wheel_slip,
        "speed": speed
    })
    
    return df, pothole_centers

# -----------------------------
# 2️⃣ DBSCAN clustering to detect potholes
# -----------------------------
def detect_potholes(df, eps=0.6, min_samples=20):
    features = df[["latitude", "longitude", "iri", "wheel_slip", "speed"]]
    scaler = StandardScaler()
    X_scaled = scaler.fit_transform(features)
    
    db = DBSCAN(eps=eps, min_samples=min_samples)
    df["cluster"] = db.fit_predict(X_scaled)
    
    return df

# -----------------------------
# 3️⃣ Estimate pothole severity
# -----------------------------
def estimate_severity(df):
    clusters = df[df["cluster"] != -1].groupby("cluster").agg({
        "latitude": "mean",
        "longitude": "mean",
        "iri": "mean",
        "wheel_slip": "mean",
        "speed": "mean",
        "cluster": "count"
    }).rename(columns={"cluster": "n_points"}).reset_index()
    
    clusters["severity_score"] = (
        clusters["iri"]*0.5 + 
        clusters["wheel_slip"]*5 + 
        (40 - clusters["speed"])*0.2
    )
    
    return clusters

# -----------------------------
# 4️⃣ Run simulation
# -----------------------------
df, true_centers = generate_vehicle_data()
df_clustered = detect_potholes(df)
severity_df = estimate_severity(df_clustered)

print("Detected Pothole Clusters:")
severity_df.head(10)

# -----------------------------
# 5️⃣ Visualization
# -----------------------------
# plt.figure(figsize=(10,6))
# plt.scatter(df_clustered["longitude"], df_clustered["latitude"], 
#             c=df_clustered["cluster"], cmap="tab10", s=15)
# plt.title("Pothole Cluster Detection with DBSCAN")
# plt.xlabel("Longitude")
# plt.ylabel("Latitude")
# plt.show()

Detected Pothole Clusters:

	cluster	latitude	longitude	iri	wheel_slip	speed	n_points	severity_score
0	0	42.331292	-83.045791	1.493765	0.019285	35.267299	1434	1.789849
1	1	42.340917	-83.100228	4.528992	0.119044	17.762733	58	7.307168
2	2	42.307043	-83.110333	4.518389	0.122719	18.426519	42	7.187488
3	3	42.316837	-83.071748	4.240866	0.116535	18.057337	46	7.091639
4	4	42.345228	-83.019673	4.357173	0.121590	18.173651	46	7.151804

Pothole Detection Simulation¶

Pothole Visualization¶