学位論文 ロバスト反トルク推定オブザーバに基づくロボット鉗子の把持力の推定

塚本, 祐介  ,  TSUKAMOTO, Yusuke

pp.1 - 66 , 2015-03-24 , 法政大学大学院理工学研究科
内容記述
Abstract— In this paper, the estimation of the grasping torque of robotic forceps without the use of a force/torque sensor is discussed. To estimate the grasping torque when the robotic forceps driven by a rotary motor with a reduction gear grasps an object, a novel robust reaction torque observer is proposed. In the case where a conventional reaction force/torque observer is applied, the estimated torque includes not only the grasping torque, namely the reaction torque, but also the estimation error torque, which is a combination of the load torque caused by the reduction gear and the load torque caused by the component parts of the robotic forceps. In this paper, a new reaction force/torque observer is proposed, which is termed robust reaction torque observer. The robust reaction torque observer is developed according to the following steps. First, cancellation of the estimation error torque, which causes an overshoot during the torque estimation, is performed using the same type of DC motor without a reduction gear. Second, identification of the load torque caused by the reduction gear is conducted in offline experiments and its cancellation is performed. Third, identification of the load torque caused by the component parts of the robotic forceps is conducted in offline experiments and its cancellation is performed. Thus, the grasping torque added to the grasping component of the robotic forceps (the gripper) is derived by subtracting the estimation error torque and the load torques from the estimated torque. Finally, the mass of the load added to the gripper is calculated from the estimated grasping torque. The proposed robust reaction torque observer was experimentally evaluated. From the experimental results, the mass of a 50 g load was able to be estimated to an accuracy of 6 g.
本文を読む

http://repo.lib.hosei.ac.jp/bitstream/10114/11734/1/14_thesis_master_%e5%a1%9a%e6%9c%ac%e7%a5%90%e4%bb%8b.pdf

このアイテムのアクセス数:  回

その他の情報