Evolving Planning Behaviors in an Artificial Ecosystem

Abstract

A tradeoff between reactive and deliberative planning exists in artificial and natural ecosystems. We introduce an evolutionary approach to balance planning behavior for agents in competitive ecosystems. Agents are oriented towards their goal by forces which are designed in analogy to smell distribution in nature. Successful agents can have mutated offspring. First, we explain our model and then we observe for different environments how the balance of deliberation and reactiveness evolves in our simulations. It can be seen that evolution adapts the planning behavior of agents successfully.