E.S.C.Ai.P.E Trailer

E.S.C.Ai.P.E. is a puzzle based on pathfinding and state machine with slight implementation of behaviour pattern and an attempt at machine learning. The main objective is to finish the level before the siren rises all the way to the top of the screen. To finish each, level, you’ll have to collect items and find pathway to create an opening to the Ai’s freedom.

This game written in C++ with an inhouse engine

The most affluential technique I used is path finding. Using Dijkstra algorithm learned in class, I achieved a good start at the movement I wanted for my AI. I automated the way the co-ordinates and its connections are created to simplify my workload as well as give me the opportunity to create dynamic map changes on the fly. I also separated the type of tiles in ways that would interact with the Ai in the game. The enemies in the game use state machine to change between different states as does the Ai player, though it is only to change from idle to moving and vice versa. The enemy can patrol a small area, chase the player if he comes to close and idle out the rest. I did not implement an attack as it I didn’t find it that important for this kind of game.

The Ai has some behaviour patterns. While I didn’t use velocity and acceleration, my fleeing action is sustained by recounting the past vectors the Ai has walked on and then walking back in case of danger. Similarly, if it sees an item, it will automatically path it’s way to the item should be able to get it. There is a line of sight at play with limited vision. Line of sight accounts for wall and any other obstacles there might be so an item or enemy beyond the wall cannot be seen.