BlockDrop AI

About BlockDrop AI

BlockDrop AI was my experiment in AI development, initially aiming to build a simple heuristic AI and later exploring machine learning techniques. The goal was to see if I could program an AI to play a block-dropping puzzle game effectively while experimenting with different approaches to AI decision-making.

• 🧠 Heuristic-Based AI – Uses a rule-based approach to optimize game moves.
• 🎓 Early Machine Learning – Implemented AI that learns but struggled with saving knowledge.
• 🕹 Classic Puzzle Mechanics – Inspired by retro block-dropping games.

AI Demo

Challenges & Lessons Learned

While I successfully created a heuristic AI, I ran into issues when implementing machine learning. The AI struggled with saving learned data and visualizing its learning process, making it difficult to debug. However, this project taught me a lot about balancing AI parameters.

• 🛠 Machine Learning Challenges – Saving learned data and understanding AI decision-making.
• ⚖ AI Optimization – Learning when to try new moves vs. sticking to proven strategies.
• 📊 Parameter Balancing – Fine-tuning decision-making thresholds for AI efficiency.

Technical Details

BlockDrop AI was built with the goal of experimenting with AI-driven gameplay in a puzzle-based environment.

• 🔢 AI Type: Heuristic-based AI with early ML implementation.
• ⚙ Algorithms Used: Rule-based decision-making, weighted scoring system.
• 💾 Machine Learning Attempt: Early reinforcement learning, but faced challenges in saving progress.

Try BlockDrop_AI

Check out the source code on GitHub or try the app (if available).