Multilayered Evolutionary Architecture for Behaviour Arbitration in Cognitive Agents

In this work, an hybrid, self-configurable, multilayered and evolutionary subsumption architecture for cognitive agents is developed. Each layer of the multilayered architecture is modeled by one different Machine Learning System (MLS) based on bio-inspired techniques such as Extended Classifier Systems (XCS), Artificial Immune Systems (AIS), Neuro Connectionist Q-Learning (NQL) and Learning Classifier Systems (LCS) among others. In this research an evolutionary mechanism based on Gene Expression Programming (GEP) to self-configure the behaviour arbitration between layers is suggested. In addition, a co-evolutionary mechanism to evolve behaviours in an independent and parallel fashion is used. The proposed approach was tested in an animat environment using a multi-agent platform and it exhibited several learning capabilities and emergent properties for self-configuring internal agent’s architecture. Index Terms — Bio-inspired Machine Learning, Gene Expression Programming, Hybrid Behaviour Co-evolution, Subsumption Architecture.