An action selection model having motivational mechanism inspired by the real rat's behaviors for autonomous robots

摘要

One of the major problems on autonomous robot researches is to select appropriate actions in unknown and changing environments. In general, an autonomous robot selects its action by priority-based arbitration. However, it is difficult that the designer prepares action's priorities for all situations in advance. Therefore, the robots should have an adaptive action selection mechanism. We focus on how animals execute their behavior in the environment. Animals behave based on their motivational mechanism, and adapt to the environments by individual experiences and social interactions. In this paper, we propose an autonomous agent model having a motivational mechanism inspired by animals. The agent takes its action by using sensory input, internal motivation and social cue from the other agent like animals. We confirm the validity of the proposed model by comparing agent's behaviors in computer simulations with the results of the animal experiment [1] which showed the social transmission among rats.