Self-Management Technique for Adaptive Robot Software based on Task Environment Similarity

In this paper, we propose a novel method by which the robot can choose the most suitable software module for the given task based on the evaluation of task environment, and the resulting relationship between the environmental information and robot software module is managed by the robot itself. In addition, the performance of the robot can be improved through the update process of the E-S(Environment-S/W module) relationship information when the new environmental information or new robot software modules are given. The effectiveness of the proposed self-management technique is shown by the experiments with 70 random maps, which shows the improvement of performance as the task continues to proceed.