MindZone is a First Person Shooter that allows the player to control the damage they produce by controlling their attention levels.
The player is set in a trapped dungeon with no means of escaping. The dungeon is filled with skeletons and mutated creatures which spawn in a fixed interval. The objective of the game is to kill as many of these enemies as possible. The game ends once the health of the player reaches 0.
The player can move anywhere along the floor of the dungeon and has a speed of 15 units. The skeleton has a base damage of 2 units and movement speed 4 units. The creature on the other hand has a base damage of 10, walking speed of 3 units and running speed of 7 units. The skeleton has 50 units of health, and the creature has 100 units of health.
| Enemy | Damage | Health | Speed |
|---|---|---|---|
| Skeleton | 2 | 50 | 4 |
| Scorpion Creature | 10 | 100 | 3(Walk), 7(Run) |
The player is equipped with a gun with a base damage of 10 units. The gun is equipped with an enhancer that increases the damage based on the player’s attention level.
Three spawners were created to drop enemies into the dungeon. Each spawner has a unique delay between each spawn and a variance. The spawners rotate around an axis which is placed at the centre of the dungeon. This is to ensure randomness of the spawn.
At the start of the game the skeleton has a spawn probability of 0.85 and the creature has a spawn probability of 0.15. After every spawn, the probability of the creature spawning increases by 0.01 and the probability of the skeleton spawning decreases by 0.75. However, the probability of the creature spawning cannot increase 0.75.
The game was interfaced with the EEG headset using the Cortex API and the EMOTIV BCI Windows Application. The player must connect the EEG headset using Bluetooth once connected, the player must create a profile and train the in-built classifier for 2 commands – “neutral” and “attention”. The “neutral” command implies the player is relaxed and the “attention” command implies the player is attentive. This problem is now transformed into a binary classification problem.
The Cortex API and Unity EMOTIV plugin was used to receive the mental commands detected by the headset. A C# script “CortexFacade.cs” was created to abstract the relevant classes of the plugin as an aggregation following the façade design pattern improving maintainability of the software for future improvements.
The “attention” command increases the base damage of the player’s gun from 10 units to 50 units and heals the player by adding 0.01 units of health if the player’s health is not 100 units. This enables the player to defeat more enemies in a shorter time
| Command | Damage | Skeleton Bullets | Creature Bullets |
|---|---|---|---|
| neutral | 10 | 5 | 10 |
| attention | 50 | 1 | 2 |
- The Player interacts with the Mind-Zone Game
- The MindZone game reads data from the Cortex API and is connected to the EEG Headset using Bluetooth
- The Mind-Zone server saves the player data gathered from the game to Firestore for further analysis
The API and code used to integrate the Game with the EEG Headset can be found in this repository https://github.com/Emotiv/cortex-v2-example/tree/master/unity
The MindZone game uses a Facade Deign Pattern to encapsulate all the neceassry classes of the Cortex API into a single class
The game contains two different types of enemies which are controlled by different Artificial Intelligent Algorithms
The
skeleton is controlled by a Finite State Machine
(FSM) with 5 states. The initial state of the FSM is
the “idle” state, and the terminal states are the
“end” state and “death” state. The intermediate
states are the “chase” state and “attack” state. The
FSM toggles changes from the “chase” to the
“attack” state when the distance between the
player and the skeleton is less than or equal to 5
units. If the payer moves away, the state changes
back to the “chase” state. If the skeleton dies, the
FSM reaches the terminal “death” state and if the
player dies, the FSM reaches the “end” state.
The creature is controlled by a behaviour tree with 1 root node, 2 selector nodes, 1 sequence node and 8 leaf nodes.
The leaf nodes of the behavior tree are “howl”, “die”, “idle”, “walk”, “chase”, “left attack”, “right attack”, “double attack”.
The selector evaluates the conditions from left to right and the leaf node with the first successful condition is the state selected. If the game is over, then the selector chooses the “howl” node. If this condition fails and the creature’s health is 0, the selector chooses selects the “die” state.
Otherwise, if the creature has just spawned, the node chosen is the “idle” node. The selector contains another selector to evaluate. The second selector must choose between the “walk” and “chase” nodes. The creature will walk for the first 90 seconds of the game and after that it will run with a greater speed. This is to ensure that the difficulty of the game increases progressively. To make the enemy AI more intelligent, a sequence node was used to give it an attack pattern. The creature hits the player with its left arm, then with its right arm and finally with both arms. The damage for all these types of attacks is the same. Since the attack pattern can be thought of a sequence of attacks, a sequence node is best to model this behavior.
