The simulation runs autonomously. When all allies or enemies are dead, press R to restart the simulation. You can click and drag the map and zoom in/out with mouse scroll wheel.

This project explored the design and development of squad-based AI behaviours for a pre-built 2D top-down shooter as part of a university coursework submission, developed using Unity and C#. Core gameplay systems and maps were provided, allowing the work to focus entirely on developing the AI.

The AI I developed features squad-based behaviour with defined roles, formation movement using steering behaviours, finite state machine decision making, target and weapon selection, friendly-fire avoidance, and A* pathfinding blended with steering to navigate the map. Agents make context-aware decisions about when to attack, retreat, or regroup based on squad state and combat conditions.


AI features implemented

  • Squad-based behaviour
    Allied agents operate as a coordinated group with defined roles such as leader, flanker, and support.

  • Formation movement
    Allies maintain formation positions behind a leader using steering behaviours, allowing the squad to move as a unit without clumping.

  • Decision making (FSM)
    Agents use a finite state machine to decide when to follow the squad, attack enemies, or regroup.

  • Target selection
    Agents choose enemies to engage based on distance and current squad context.

  • Weapon selection
    Multiple weapons are available, with AI choosing when to use melee, guns, or rockets based on range, enemy density, and safety.

  • Friendly fire avoidance
    Agents avoid shooting when allies are in the line of fire and use additional safety checks for explosive weapons.

  • Navigation support
    A* pathfinding is used to navigate procedurally generated maps, blended with steering behaviours for smooth movement.

Leave a comment

Log in with itch.io to leave a comment.