Method of action selection of an autonomous mobile robot using the immune networks

Recently, in artificial intelligence technology, much attention has been focused on the approaches based on behavior (behavior-based AI) that demonstrate robustness and flexibility to dynamically changing environments. On the other hand, the immune system plays important roles in maintaining its own system in dynamically changing environments. Therefore, we believe that the immune system can provide a new paradigm suitable for dynamic problems dealing with unknown environments rather than static problems. We propose a new approach to behavior-based AI by focusing our attention to the biological immune system. The basic principle of our proposed method is that the immune system equipped with the robot selects the simple action (antibody) which is suitable for the current situation (antigen). We confirm the feasibility of our proposed method by applying it to the behavior control of an autonomous mobile robot in a cluttered environment as an example.