RLIDS Based Goal Improvement in Emotional Control

In this paper, a goal improvement approach based on experiences for handling external disturbances is proposed. The suggested method uses reinforcement learning by an actor-critic mechanism in order to adapt the system to disturbance and uncertainty. Goal improvement is achieved via IDS. IDS is a powerful method in information-handling processes. It has a fast convergence and simple underlying, away from complex formulas. Temporal Difference (TD) learning is utilized in updating both Stress-Evaluator and the goal of emotional controller. As a benchmark to evaluate our method, inverted pendulum and simple submarine systems are used. Results show that the proposed method responds better to external tackling disturbances in comparison with the former fixed goal model.