An agent-based simulation with a mischievous premise: a robot must keep a house clean while children wander around making dirt. The robot wins if it pens all the children and gets the house spotless; it loses if dirt ever covers 60% of the floor.
The house is a discrete grid where everything happens in turns: agents act first, then the environment changes on its own at random intervals, so the robot’s plans keep breaking. The robot sees the full state of the world and must decide, every turn, whether to chase children, carry them to the playpen, or clean, and in which order, with its strategy racing the spread of mess.
The project tests how different agent behaviors hold up under this pressure, with a written report on the strategies and their results across simulation runs.
Written in Python from scratch, including the environment, the agents, and the simulation loop.