🐺 Bio-Inspired Cognitive Swarms

Grey Wolf Optimization Augmented with Short-Horizon LSTM Prediction

---

📌 Overview

This project presents a hybrid GWO-LSTM framework for intelligent UAV swarm coordination. Traditional swarm planning methods react only to current target observations — failing when a target executes evasive maneuvers. This framework solves that by integrating LSTM-based trajectory prediction with Grey Wolf Optimization (GWO), enabling the swarm to move proactively toward an anticipated target position rather than reacting after the fact.

Architecture: Predict → Optimize → Execute → Feedback

---

🎯 Key Features

• 🔮 LSTM Prediction Module — predicts future target position using recent trajectory history
• 🐺 Grey Wolf Optimization — decentralized swarm coordination using α–β–δ–ω hierarchy
• 💡 Hybrid Coordination — predict-then-optimize pipeline for proactive interception
• ⚡ Multi-Objective Fitness — optimizes interception delay, energy, collision risk, and communication
• 🔄 Continuous Feedback Loop — adapts dynamically to target maneuvers in real time

---

🏗️ System Architecture

The framework consists of four main layers:

Layer	Description
Temporal Observation Layer	Collects target trajectory history
LSTM Prediction Layer	Predicts future target positions (h steps ahead)
Optimization Layer	GWO-based trajectory planning and role assignment
Swarm Execution Layer	Executes optimized UAV movements

---

🔧 Tech Stack

Component	Tool
Language	Python 3.x
ML Framework	TensorFlow, Keras
Numerical	NumPy, SciPy
Visualization	Matplotlib
IDE	Jupyter Notebook / VS Code
OS	Windows / macOS / Linux

---

⚙️ Simulation Parameters

Parameter	Value
Search Space	1000 × 1000 m²
UAV Max Speed	20 m/s
Target Speed	14 m/s
Communication Range	150 m
Time Step	0.5 s
GWO Population Size	30 Wolves

---

🚀 Getting Started

Prerequisites

pip install tensorflow keras numpy scipy matplotlib

Run the Simulation

# Clone the repository
git clone https://github.com/RabbaniHacker/uav-swarm-interception.git
cd uav-swarm-interception

# Launch Jupyter Notebook
jupyter notebook

# OR run directly
python main.py

---

📁 Project Structure

bio-inspired-cognitive-swarms/
├── main.py                  # Entry point
├── gwo/
│   └── optimizer.py         # Grey Wolf Optimization logic
├── lstm/
│   └── predictor.py         # LSTM trajectory prediction model
├── swarm/
│   └── coordination.py      # Swarm coordination and role assignment
├── simulation/
│   └── environment.py       # 2D environment, targets, obstacles
├── utils/
│   └── visualization.py     # Matplotlib-based plotting
├── notebooks/
│   └── demo.ipynb           # Interactive demo
└── README.md

⚠️ Adjust folder names to match your actual project structure.

---

📊 How It Works

1. Observe — UAVs collect current target position and state data
2. Predict — LSTM network predicts future target coordinates (h steps ahead)
3. Optimize — GWO generates optimal swarm trajectories toward predicted position
4. Execute — UAV swarm performs coordinated interception
5. Repeat — Updated positions feed back into the next planning cycle

---

📚 References

• Kori et al., GWO-based Cognitive Swarm Attack Model, IEEE, 2023
• Mirjalili et al., Grey Wolf Optimizer, Advances in Engineering Software, 2014
• Hochreiter & Schmidhuber, Long Short-Term Memory, Neural Computation, 1997
• Shukla et al., LSTM-based Trajectory Prediction, IEEE, 2025
• Tang et al., Receding-Horizon Evolutionary Planning, Aerospace Science and Technology, 2019

---