The Origin: Understanding the Basics (2022) Every complex system starts with a fundamental question. In 2022, my question was: How does a computer know the shortest route from Point A to Point B? To answer this, I built my first pathfinding visualizer using Python and Pygame. The original iteration was extremely rigid—a simple binary grid where squares were either open paths or impassable walls. I implemented the A* algorithm using a standard heuristic approach, visualizing how the algorithm mathematically balanced the distance already traveled with the estimated distance to the goal. It was a functional proof-of-concept, but it lacked real-world application.
The original project running A star
The Evolution: From Pixels to Physical Constraints (2023 - Present) As my interest deepened into mechatronics and artificial intelligence, I realized that real robots do not move in a frictionless binary world. After six major iterations over two years, I completely re-architected the codebase.
I transitioned the architecture from a monolithic script to a robust Object-Oriented design, utilizing a TerrainNode class to encapsulate state and a central PathfinderApp class to handle the simulation engine. This allowed me to introduce complex, real-world variables necessary for actual robotics:
- Variable Terrain Costs: I implemented a weighted graph system. Instead of just "open" or "blocked," algorithms now calculate routes across grass (cost: 1), roads (cost: 0.5), water (cost: 3.0), and mountains (cost: 5.0).
- Kinematic Constraints (Turn Penalties): In robotics, moving forward is energy-efficient, but braking, rotating the chassis, and accelerating drains battery and costs time. I modified the pathfinding algorithms to apply a mathematical penalty (+0.5 cost) every time the path forces a change in direction. The algorithm now favors smooth, continuous movements over jagged, technically shorter lines.
- Multi-Algorithm Engine: I integrated multiple algorithms (A*, Dijkstra, and Greedy Best-First Search) into a single simulation engine. This allows users to actively compare how different logic performs under the exact same terrain constraints in real-time.
The A star algorithm on the new project
The greedy algorithm for comparison
Technical Growth and Takeaways This project represents my growth from writing sequential code to designing scalable systems. It required a deep dive into data structures, specifically utilizing Priority Queues (heaps) for performance optimization during node evaluation. By adding kinematic constraints and terrain weights, I transformed a standard computer science exercise into a practical tool for mechatronics logic, proving that efficient software design is crucial for optimal physical hardware execution.