Experimental Python project to explore video game-style 'AI' behaviour in a 2D top-down environment.

Intended to focus on simulated 'Bot' entities' behaviour at an 'action'/'tactical' level, e.g. searching, spotting, assessing and attacking.

Python isn't the best tool for this, but it's largely a learning exercise.

Implemented as a library. User interface uses pygame-ce.

# Namespace may change...
from two_d_game_ai.entities import Bot
from two_d_game_ai.render.view import View
from two_d_game_ai.world import World

# WORLD

# Create a World
the_world = World(
    size=300,
    grid_size=16,  # used for route planning
)

# ENTITIES

# Add a Bot to the World...
b0 = Bot(
    name="b0",
    world=the_world,
    pos=(20, 20),
)

# ... with a destination
b1.destination = (25, -50)

# Create a View to the World.
# Provides user interface, but not required: e.g. Bot, World test suites don't have one.

view = View(
    name="the_view",
    world=the_world,
    scale_factor=2,
)

# MAIN LOOP

while view.running:
    view.handle_inputs()
    if not the_world.is_paused:
        the_world.update()
    view.render()

• Square
• Divided into a square Grid

• Circular entity that blocks Bot movement

• Treated as a point
• Initially stationary
• While stationary, can rotate at a constant rate
• Can move at a constant rate in the direction it is facing
• Can be given a destination; will plan and follow a route to it.
• Can see Bots it's facing
• Keeps track of visible and known (previously visible) other Bots

• Centered on origin (0, 0); conventional (positive, right-handed, y-axis up) coordinate system
• By default, 1 World unit : 1 display pixel; can be scaled at initialisation
• Updates initially paused, P key toggles pause
• Renders every update at up to 60 updates/second; is allowed to lag, i.e. may run slower than real-time.

• As icons with:
  • Direction indicator
  • Destination point and waypoints on calculated route to it
  • Vision cone
  • Can-see and knows-about relationships as lines
• Primary mouse click to select a Bot; secondary mouse click to set a destination

• Invoke can be used for CLI tasks (autoformatting, linting and type checking). invoke --list or inv -l to get a list of tasks.