脚部に重力補償機構を組み込んだヒューマノイドロボットの設計開発評価

Abstract

本論文では、以下の章から構成される。第1章「序論」では、本研究の背景および目的について述べる。第2章「受動テンドン型重力補償機構」では、はじめにヒューマノイドロボットにおける重力補償機構の必要性について議論する。次に、今回提案する受動テンドン型重力補償機構の設計を行う。重力補償機構は膝関節用と股関節・足首関節用の2種類を設計する。そして、ヒューマノイドロボットHRP-2の実験結果をもとに、もし、HRP-2にこの重力補償機構を搭載した場合に、モータトルクにどのような変化が見られるかを計算によって求め、重力補償機構の効果を確認する。また、重力補償機構を設計するために必要なパラメータを選定する。第3章「重力補償機構を組み込んだヒューマノイドロボットの設計および開発」では、第2章で設計した重力補償機構を実際に組み込んだヒューマノイドロボット才華4の脚部と胴体部の設計と開発を行う。ヒューマノイドロボット才華4は、受動テンドン型重力補償機構を組み込んだヒューマノイドロボットという側面と、現在使用している現行機である才華3の後継機であるという2つの側面をもつ。まずはじめに、才華3の問題点を踏まえたうえで、才華4の設計方針を示す。次に、設計方針に合わせて実際に行った脚部と胴体部の設計を行う。特に、脚部の設計では、第2章で設計した重力補償機構を実装する上での、変更点などについて述べる。最後に、搭載する制御用デバイスや制御プログラムなど、制御システムについて述べる。第4章「評価実験」では、第3章で設計、開発したヒューマノイドロボットを用いて、屈伸動作と足踏み動作といった簡単な動作実験を行う。複数の条件で実験を行い、それらの実験結果を比較し、考察する。関節角度誤差から、モータのトルクと出力を推定し、それを用いて、重力補償機構の実際の効果を確認、評価する。第5章「結論」では、以上の議論を要約し、結論を述べる。また、本研究により明らかになった課題と展望についても述べる。

The objective of this thesis is to design and develop a humanoid robot equipped with a gravity compensation mechanism in the legs, and to evaluate a gravity compensation mechanism. There are three concrete objectives as follows: (1) Design of a passive tendon type gravity compensation mechanism. (2) Design and development of a biped humanoid robot equipped with a gravity compensation mechanism. (3) Verification of effectiveness and evaluation of a gravity compensation mechanism. First, a gravity compensation mechanism is introduced. The mechanism proposed in this thesis is based on a passive tendon mechanism using springs to produce recovery force. The compensation ratio against the gravitational torques working on the motors of the supporting leg is constant, unaffected by the joint angles. The design is simply composed of springs, wires, and pulleys, making it possible to apply on humanoid robots in general. Using the humanoid robot HRP-2, number of experiment to examine the effectiveness of the mechanism is performed. The experiment was carried out by evaluating and comparing the torques at joints. Motions of the experiment are knee bent standing on the ground (assuming the support phase), Knee bent hanging in mid-air (assuming the swing phase), and dynamic walking. The differences in the torques with and without the gravity compensation mechanism are computed. From this analysis, the effectiveness of the gravity compensation mechanism in reducing gravitational torques is validated along with the optimum compensation ratio. Second, Humanoid robot Saika-4 is designed and developed which is equipped with a gravity compensation mechanisms in the legs. The Saika-4 is a self-sustaining human size humanoid robot. In order not to give human fear, the height of the Saika-4 is set to 1,500 mm or less, and the weight is set to 60 kg or less. The size of each part of the Saika-4 refers to a youth males size. Thereby, it is made the form to which human does not have sense of incongruity in the Saika-4. The Saika-4 will have 6 DOF (Degree Of Freedom) at each leg, 7 DOF at each arm, 1 DOF at each hand and 2 DOF at robotics head, consequently, 30 DOF in total. The important development policies are equipment of the gravity compensation mechanism and improvement of maintenance efficiency. The gravity compensation mechanism is applied to the legs of the Saika-4. The structure of mechanism for the Saika-4 is designed based on the compensation ratio (it is 0.5), the weight of upper body and the length of legs. By standardizing and simplifying the structure at the joints and also at other parts, the number and variety of the components was greatly reduced. Therefore, installation of the Saika-4 is easy, a higher maintenance efficiency is obtained, and reduced manufacturing costs is possible. Moreover, because the Saika-4 is composed of several modules, installation is easy and maintenance efficiency is increased. The Saika-4 is equipped with a PC, batteries, a wireless Ethernet modem, a gyroscope, motor drivers in the body, which are electronics device required for self-sustaining robot. Now, development of the legs and a body is completed. The height of present state (from sole to shoulder) is 1,225 mm, and weight is 44 kg. Finally, using the humanoid robot Saika-4, it is performed an effectiveness evaluation of the gravity compensation mechanism. Observations are performed, of the robot with knee bent standing on the ground, knee bent hanging in mid-air, and during a stamping motion. The stamping motion is performed to evaluate the effect of the support phases and the negative effect of swing phases collectively. Because this mechanism is meant to compensate gravitational torques during steady state, the motions implemented were performed in low speed. By comparing the torques with and without the compensation mechanism applied, the effectiveness of the mechanism could be evaluated. The evaluation was performed using the absolute mean value of the torques and electric power, estimated from errors in the joint angles, along with other indicators. From this experiment, the effectiveness of the gravity compensation mechanism is validated. In particular, the effects of this mechanism is especially large on the knee joint, with a decrease of gravitational torques up to about fifty percent during knee bent standing on the ground, which is close to initial predictions. The joint torques during knee bent hanging in the mid-air with mechanism are almost the same as the joint torques during knee bent standing on the ground without mechanism. Therefore, negative effect of swing phases of the gravity compensation mechanism is small. In addition, about ten percent decrease during stamping motion is observed at all joint.