🔥 Dynamic Patrol Optimization in Unknown Continuous Space

This project simulates a dynamic patrolling system in an unknown 2D continuous space where the goal is to:

1. Discover hot points (areas of interest or activity)
2. Find the shortest closed path to compare with patrol path
3. Dynamically patrol these points to minimize the time they are left unvisited

---

📌 Project Structure

The experiment is divided into three main phases:

🧭 Phase 1: Hot Point Discovery (Exploration)

• Space is continuous and initially unknown.
• A synthetic "heat map" is simulated using Gaussian peaks.
• Particle Swarm Optimization (PSO) is used to discover the "hot" points (local maxima) in this space.

🧠 Phase 2: Fastest Path

• Once hot points are discovered, we compute the shortest closed path visiting all of them (TSP).
• This is solved using Ant Colony Optimization (ACO) to simulate cooperative path building.

🚨 Phase 3: Dynamic Patrolling

• A single agent patrols the discovered hot points.
• Instead of following a fixed loop, the agent uses a dynamic scoring strategy to prioritize:
  • Time since last visit
  • Distance to each hot point
  • (Optional) Importance of the point
• The goal is to minimize the max time between visits to any hot point.

---

📊 Extensions

This test is extensible to:

• 🔁 Multiple patrolling agents
• 🧠 Reinforcement Learning-based patrol logic
• ⚡ Energy-aware or probabilistic patrols
• 🎯 Adaptive importance weighting

📽️ Animation & Visualization

• Full visualizations of the heat map, discovered hot points, patrol path, and visit intervals.
• Animated patrol showing the latest movement segment only for clarity.

---

📦 Dependencies

pip install -r requirements.txt