Journal of the Robotics Society of Japan

This paper focuses on the feedforward position control for a musculoskeletal system which has two links, two joints and six muscles. This feedforward control method utilizes resultant torque generated by inputting the internal force among muscles balancing at a desired posture. The motion convergence of this control method is extremely sensitive to the muscular arrangement. Using the approximation of Taylor expansion, the previous study clarified the sufficient condition of the muscular lengths for converging at a desired posture. This paper expands this discussion to clarify the geometric conditions of the muscular arrangement.

View full abstract

Show abstractHide abstract

In this paper, a new set-point control method for a musculoskeletal system by combining a feedback control including a large time-delay with a muscular internal force feedforward control is proposed to accomplish a robust and rapid positioning with relatively low muscular contraction forces. It is known that feedback control under the existence of a large time-delay owns a drawback that there is a serious trade-off between a control performance and its stability. On the other hand, the muscular internal force feedforward control, which we have already proposed, also owns a drawback that it requires a large muscular contraction forces to make a good performance in a wide range. A simple linear combination of these two controllers makes it possible to improve the control performance and to complement each drawbacks each other. Firstly a two-link six-muscle arm model is given as a specific musculoskeletal system treated in this study. Secondly the new control law, which is composed of the feedback control signal including the time-delay and the internal force feedforward control signal, is constructed by only using a kinematic information of the musculoskeletal arm model. Next, a stability of the proposed method is discussed using Lyapunov-Razumikhin method. After that, results of several numerical simulations are shown to demonstrate several advantages of the proposed method.

View full abstract

Download PDF (2267K)

For a shooting robot capable of launching a rotating object with an angular velocity of ω and eventually landing at the pre-determined point, we first discuss the issue on how many actuators and sensors are necessary and sufficient by utilizing. Next, we show that a linear actuator can excite the rotation of object, release it by using an appropriate trigger mechanism, and wind the tether. We also show that just one position sensor is enough for determining the releasing angle θ as well as ω, with the assistance of disturbance observer. Finally, we design and develop the robot system to confirm the idea through experiments.

View full abstract

Register with J-STAGE for free!