Pioneer Pathfinder - Brad Flavor
A greedy pathfinding algorithm with some optimizations for f-score tie-breaking and using a octile Chebyshev distance to estimate h-costs. Iterative, stepped implementation with vanilla JS class OOP, in a 16,384-cell HTML table. Obstacles are based on primes, inflated to 4 or 16 surrounds.
Click a visited cell to see its data.
The thing that makes this different from A* pathfinding is that A* has to explore the entire map first to know its final path, working backwards. This one doesn't. It does its best with a limited field of view, but with more invested in estimation than a typical greedy algoritm, making it more accurate. Consider a "pioneer" like a self-driving car or land robot that knows the coordinates of its target, but hasn't yet had the opportunity to explore the obstacles much further away. I chose to make it more realistic, like for a land-navigating subject with limited-distance sensors, as opposed to an idealistic analysis, like a what obstacles a bird or plane might sense, and then going back to trace a path. It approximates a globally optimal solution through good local estimation. I'll be using it for a self-driving RC car project.