Drone swarm simulation
A Master’s thesis project combining robotics, evolutionary algorithms, and physics-based 3D simulation. The system studies a swarm of 5 UAVs planning routes through complex environments (forest, urban) using PyBullet physics and a virtual LiDAR sensor. Routes are generated by four bio-inspired heuristic algorithms (MSFOA, OOA, SSA) alongside classical NSGA-III as a baseline, while in flight the drones reactively avoid obstacles through real-time local path replanning. The entire pipeline — from simulation to statistical analysis and automated reporting — is designed for full experiment reproducibility (240 processed runs). Working on this project let me combine engineering craft with scientific rigor: from a Strategy-pattern-based architecture, through implementing multi-objective metaheuristics, to statistically validating the results.
- Python
- PyBullet
- gym-pybullet-drones
- NumPy
- SciPy
- Pandas
- Matplotlib
- Hydra
- h5py
- SQLite
- Jupyter
- Pytest
- Mypy