非線形弾性腱を持つ腱駆動ロボット機構の定力制御

抄録

Force control of a tendon-driven robotic mechanism is usually done with an active compliance controller, even if the mechanism has passive compliance elements. The controller usually consists of multi feedback loops where the inner loop controls the tension forces of tendons, and the outer one controls the joint angles of the robot. For such a controller, it is assumed that the inner loop converges quickly enough so that we can ignore the dynamics of the inner loop when we design the outer feedback loop. This turns to restrict the bandwidth of the outer loop. Also it means that the stability is not proved strictly.
In this paper, we show that a tendon-driven mechanism with nonlinear tendon elasticity should be controlled in the actuator level in order to utilize the performance of the compliant elements efficiently. As one of such controllers, a set-point PD controller is applied to a tendon-driven robotic mechanism to control the contact force with the environment and the Lyapunov stability is investigated. The stability analysis of a tendon-driven manipulator is rather complicated than that of a conventional manipulator or an elastic joint manipulator, because a tendon-driven manipulator has more numbers of tendons than that of the joints and the tendons have nonlinear elasticity. We derive a set of sufficient conditions for the Lyapunov stability and show some numerical simulation results.