Model-based reinforcement learning approach for deformable linear object manipulation

摘要

Deformable Linear Object (DLO) manipulation has wide application in industry and in daily life. Conventionally, it is difficult for a robot to manipulate a DLO to achieve the target configuration due to the absence of the universal model that specifies the DLO regardless of the material and environment. Since the state variable of a DLO can be very high dimensional, identifying such a model may require a huge number of samples. Thus, model-based planning of DLO manipulation would be impractical and unreasonable. In this paper, we explore another approach based on reinforcement learning. To this end, our approach is to apply a sample-efficient model-based reinforcement learning method, so-called PILCO [1], to resolve the high dimensional planning problem of DLO manipulation with a reasonable number of samples. To investigate the effectiveness of our approach, we developed an experimental setup with a dual-arm industrial robot and multiple sensors. Then, we conducted experiments to show that our approach is efficient by performing a DLO manipulation task.