バイラテラル制御に基づく模倣学習における双腕協調動作のための学習モデルの検証

抄録

Robots are expected to replace the household chores and manual labor performed by humans. However, object manipulation by a dual-arm robot is still difficult because it is necessary to consider interaction between the dual arms. Therefore, in this study, we extend imitation learning based on bilateral control, which is fast and highly adaptive to environmental changes, to the cooperative motion of dual-arm robots. In addition, we verify a model for learning the cooperative motion of the dual arms. From the experimental results, we find that the model for learning the cooperative motion of the dual arms requires the information of the dual arms to be input to a single predictor. In particular, we conclude that a model in which each arm's information is processed by a separate predictor and then integrated is effective.