SMART ENEMY AI

The idea is around an FPS game where the player has to secure a package from a guarded facility. In order to do that the player has to cross a series of hurdles and enemies, to reach the position and grab the package or interact with it. It’s a stealth game where the goal is to complete the objective with as little contact as possible. There are multiple ways to achieve a single goal and hence could have variations of approach. 

Enemy AI

The concept of this AI is around the consciousness of how they learn through the actions and events around them. 

Proposal is to create an AI which records the movement and actions of player when once the player plays that level and if player comes again to that position to play it again then the enemy would have a sense of how player is going to respond because of the past experience so they will present the hurdles in other ways. 

 

  1. Enemy can position itself according to the past movement, every time a player restarts the level.

  2. The enemy would be more attentive to a site on map from where the break in happened and would try to cover possible paths originating from it.

  3. Setting alarm systems according to the past player movement.

  4. The enemy would follow a different approach of combat according to the player's game style. It's not necessary for the enemy to guard or hunt every time, it can use a tactical method to achieve the goal.

Security System AI

The AI is based on the sensory readings where the objects in the game like camera and alarms read the player actions and try to counter the style of the player by giving new placement and growing intelligence towards the player.

 

  1. Like the rotation period of the camera could be different from what you played last time. 

  2. The alarm setting can be different from last time and based on the usual path you take. 

An application for the A* algorithm

The concept is based on an Air battle (3D Space) between players controlling drones and playing against each other to win. This game is based on a barren planet where there is no life but two other planets are fighting to get control over the planet. 

THE FIGHT OF DRONES

In the game play two players play against each other where they deploy their drones and try to kill each other by funding them in the map and killing them by using different sets of powers. 

Normal fire power, medium power up and super power up. 

 

  1. So the idea is about a super power up evading technique. The super power is a coordinate seeking missile which tracks down the enemy coordinate when once locked and kills on contact.

  2. The player in such a case can deploy a mini device to fake it’s coordinate for a brief time to get the time to run from the missile. 

  3. The super power missile uses the A* Algo for tracking the enemy target.

  4. The enemy deploys a mini device to fake the location the missile updates the target location and resets it after the mini device defuses. 

  5. In this case as long as the superpower is on, it tracks down the enemy location. 

  6. The 3D space makes the use of A* Algo more interesting and dynamic due to the change in target location.

DIJKSTRA ALGORITHM Vs A* ALGORITHM

The idea between these two algorithms could be understood by a simple example of combat conditions where you have to assassinate a boss and you are in combat with the boss and his accomplices.

 

While combating, in use of guns there is an algorithm in work which helps you to auto lock the target you are shooting. The auto locking system works on the Dijkstra algorithm. The closest enemy to your pointer in your view is the one it locks your aim to. After the enemy is dead it goes to the next closest to the pointer in view. 

 

Where as for the main boss the arrow or any signal sign to show his position on the screen is done through A* algorithm. The main boss is indicated through an arrow or a danger shade in your view so you be vigilant about him while fighting with others. 

THOUGHT PROCESS

AI.png