| Summary: | Real-life ecosystems are extremely complex, being affected by a wide variety of dynamic variables such as changes in climate, availability of food, predator-prey ratio and other variables. Simulating an ecosystem is difficult, given the many factors that interact and animals that coexist. Many ecosystem simulations exist already, however many of them lack procedural terrain generation. Hence, we aim to create an ecosystem simulation which models natural selection in an ecosystem with competing agents in a procedurally generated environment. All components in this simulation was created with the Unity Game Engine. The agents within the ecosystem are modelled to be simplified but accurate versions of their real-life counterparts. We place AI agents in a simulated environment, and analyse how different agents evolve and adapt to each other and their surroundings. We vary the initial environmental and agent parameters and repeatedly run the simulation to investigate whether an equilibrium can be reached given the initial parameters. Running these simulations show that reaching equilibrium is feasible even with random procedurally-generated terrain, as long as suitable initial parameters are chosen.
|
|---|