This project implements a top-down simulation of a differential-drive robot that learns to navigate and avoid obstacles using the NEAT (NeuroEvolution of Augmenting Topologies) algorithm. Developed as part of the Undergraduate Summer Research Fellowship (USRF) at Amity University, Noida, this work explores the integration of evolutionary computation and real-time robotics-inspired simulations.
- 🚗 Differential Drive Control
- 🧭 7 Virtual Distance Sensors for obstacle detection
- 🌱 Evolutionary Learning with NEAT using adaptive fitness
- 🧩 Randomized Obstacle Environments for generalization
- 📊 Live Neural Network Visualization
- 📈 Fitness Logging & Graphing for performance tracking
- Python 3.x
- NEAT-Python
- Pygame
- PyMunk
- NumPy, Matplotlib
The robot's fitness is determined by:
- Distance covered (exploration)
- Speed of motion
- Avoidance of collisions
A robot is rewarded for covering new ground without hitting obstacles, making the evolution goal-safe and efficient navigation.
Abdul Malik
USRF Research Fellow – Amity University, Noida
B.Tech Computer Science, Integral University, Lucknow
📍 India
🔗 LinkedIn
📧 abdulmalik256786@gmail.com
🌐 Portfolio
- Amity University, Noida — Amity Centre for artificial Intelligence
- Under the Guidence of Dr. Abhishek Kaushal and Dr. M.K Dutta
- NEAT-Python by CodeReclaimers
- Pygame + PyMunk for 2D physics simulation
- USRF Summer 2025 Research Fellowship