Our project was developed for our client, Lockheed Martin, to help address the challenge of modeling enemy behavior. We wanted this behavior to take place in a tactical PvE(Player vs Entity) game where enemy behavior could quickly adaptative to keep it unpredictable and continue to challenge the player.
In many modern games players can quickly discover a dominate or meta strategy and use it repetitively which can cause the game to lose its difficulty. Traditional difficulty systems often try and fail to fix this issue, choosing to instead artificially inflate enemy stats rather than encouraging players to adapt new strategies. The goal of this project is to explore adversarial gameplay modeling by creating a system where the game's mechanics create a way for the game to dynamically respond to player decisions. We worked in collaboration with a graduate team tasked with developing the core AI model, while our primary goal is to build the game environment required to support that system.
To achieve this goal we are building a turn based tactical mech combat game inspired by games like Into the Breach using a custom C++ backend(GDExtension) paired with a Godot Engine frontend. The game is played on an 8x8 grid map featuring unique terrain elements, and places player controlled mechs against waves of AI enemies, where the player also has to defend multiple cities from the AI. The core gameplay is controlled by our TurnManager where enemies first ready their movement and attack actions, this forces the player to respond by using their mechs to mitigate damage and control the battlefield using elements such as terrain effects and knockback. Our game includes multiple specialized mechs, a dynamic DifficultyEngine to manage encounter parameters, and a RunManager with a shop for between match upgrades. By having a tactical gameplay experience combined with distinct gameplay states and a strict action validator our game functions as both a standalone strategy game and an interactive sandbox for the graduate teams model.
