Agent Evacuation Simulation

Created for my Final Year Project focusing on the combination of different AI algorithms, this programme simulates the evacuation of up to 50 AI agents from a building that is burning.

Created Using Unreal Engine 4.25

Simulation Rules:

There are 5 types of agents with different pathfinding capabilities:

• Leaders use a "Distance to Exit" heuristic to select rooms nearest to the floors exit. Other Agents will follow leaders if they are nearby instead of using their own pathfinding.
• Employees pathfind in the same way, but will not attract followers.
• Visitors will select random neighbouring rooms, but will not enter previously visited rooms. If all nearby rooms have been visited, they will select a random room on their floor.
• Lone Wolves behave the same way, but will not follow leaders.
• Panickers select random neighbouring rooms, but do not remember previously visited rooms.

Agents must escape the building in an amount of time which can be set by the user. Every 30 seconds, new fires will spawn and some will be extinguished. The amount of agents for each simulation can be customised before starting, with up to 50 agents in total.

Controls:

Mouse Only.