RocketMind
The Project
RocketMind was inspired by my fascination with SpaceX’s autonomous rocket landings. Watching real Falcon 9 boosters perform precise, intelligent landings sparked a curiosity about the physics and decision-making behind them. I wanted to simulate that process in a way that wasn’t just educational but also interactive, something that could bridge machine learning, real-world physics, and game design. I realized there weren’t many games or platforms where players could explore how trained AI models make landing decisions, so I built RocketMind to fill that gap, where players can test, visualize, and learn from a trained reinforcement learning model in action. Simulating real-world rocket dynamics in Unity was one of the biggest challenges. I had to model physics like mass changes due to fuel burn, variable thrust, and environmental forces—all while ensuring it ran efficiently in a game engine. Another challenge was making the trained AI model both accurate and interpretable for players; I had to strike a balance between realism and playability.
