Search

In this paper, a new robust force controller is proposed for Series Elastic Actuators (SEAs) by using conventional Sliding Mode Control (SMC) and Disturbance Observer (DOb). The position measurement of the actuator's link is subtracted from the desired deflection of the spring so that the force control goal is defined as the desired position of the motor; i.e., the force control is simply performed by designing a position controller at motor side. However, the position accuracy of the motor is influenced by the dynamics of the actuator and environment. Since they cannot be easily identified in practice, the dynamics of the actuator's link and environment are considered as unknown disturbances in the design of the proposed controller. In order to improve the robustness, conventional SMC-based robust force controller is designed without considering the unknown disturbances. Although the robust force control can be theoretically performed by using the conventional SMC-based controller, it suffers from chattering in real implementations. Conventional DOb-based robust motion controller is designed at motor side so that not only the disturbances are cancelled by feeding-back their estimations but also the control signal of SMC-based robust force controller is lowered. Hence, a simple yet efficient robust force controller is designed for SEAs. The validity of the proposed robust force controller is verified by giving experimental results.

In this paper, a novel robust position controller is proposed for two-mass resonant systems. It is designed by using Differential Flatness (DF) and Disturbance Observer (DOb) in state space. Firstly, DF-based trajectory tracking controller is designed by neglecting plant uncertainties and external disturbances. Since a two-mass resonant system is controllable, state and control input references are generated in terms of differentially flat output. However, the trajectory tracking controller is sensitive so the stability and performance may significantly change due to the robustness issues in practice. Secondly, the robustness is achieved by treating estimated disturbances, which are obtained via DOb, in the design of the controller. A two-mass resonant system includes matched and mismatched disturbances; therefore, the robustness cannot be achieved by directly feeding-back their estimations. The states of the system are re-constructed by using estimated disturbances so that the mismatched disturbance is automatically cancelled via state feed-back control. The matched disturbance is simply cancelled by feeding-back its estimation. Hence, the robust trajectory tracking controller is designed for two-mass resonant systems. The validity of the proposal is verified by giving simulation results.

This paper proposes a new state-space controller for the robust trajectory tracking control problem of multi-mass resonant systems. It is designed in state-space by combining Differential Flatness (DF) and a higher order Disturbance Observer (DOb). The former systematically generates the control input and state references of a state feed-back controller so that not only regulation but also trajectory tracking control can be performed in state-space. They are obtained in terms of differentially flat output variable in DF. The latter estimates disturbances and their successive time derivatives. Its order depends on the resonant modes of the system. The proposed controller has a two-degrees-of-freedom motion control structure. Its performance and robustness can be independently adjusted by tuning the state feed-back controller and DOb, respectively. It is designed in two steps. In the first one, the state feed-back controller is tuned by only considering the nominal plant model of the system. The nominal performance is adjusted by using pole placement; e.g., all poles of the nominal system are placed on the real axis so as to suppress the vibration at tip. The state and control input references of the state feed-back controller are obtained by using DF. In the second step, the robustness of the system is achieved by reconstructing the state vector via the estimations of disturbances and their successive time derivatives. Thanks to the proposed new state variables, mismatched disturbances are automatically cancelled by the state feed-back controller. The matched disturbances are cancelled by feeding-back their estimations through control input. Hence, the robustness of the multi-mass resonant systems is achieved. Without losing generality, the proposal is verified by considering three-mass resonant systems. Its simulation results are given to validate the proposal.

Accdepted: 2006-06-13