David Hestelow

Made as a first-year project with the goal of making a recreation of the game Pac-Man within your own unique style. Due to the special constraints of the Pac-Man maze design, the generator contained some unique challenges, such as all inside walls needing to be at least 2 thick. As well as this the use of different assets based on whether the tile bordered the edge of the map or a path further complicated development.

Within the sheep AI one main challenge, especially on larger maps, was creating an efficient and accurate navigation system that would be more precise than just choosing the direction to move at each intersection. A* pathfinding was implemented to solve this, having the condition that it would only calculate a path up to 40 tiles away, instead of right to the goal, in order to get optimal performance with reliable accuracy.

This project was specially selected for and shown at UTS during

Evolutuion of the maze generator

Initial carving using just one path

Fully implemented path carving using recursive backtracking

Inital testing of correct asset placement

Further refined asset placement and path carving

Completed maze with correct asset placement and orientation