LunarPilot: DQN Fine-Tuning & Live Comparison Platform

An end-to-end Deep Reinforcement Learning project that fine-tunes a Deep Q-Network (DQN) agent to autonomously land a spacecraft in the LunarLander environment. Fine-tuning raised the landing success rate from 20% to 85% and the average reward from 157 to 226+, and a Flask + Chart.js dashboard streams both models live, side by side, for real-time comparison.