상보 필터 기반 가속도 추정기법을 통한 탄성구동기의 강인한 임피던스 제어

Abstract

This paper proposes a novel acceleration estimation method to improve the stability and performance of robust elastic structure preserving control for series elastic actuators. Elastic structure preserving (ESP) control has been proposed to overcome an inherent limitation in rendering high impedance while maintaining passivity. However, the effectiveness of ESP is highly dependent on accurate acceleration data, which is a challenge for existing estimation techniques. We introduce an acceleration estimation method that fuses position sensor data with a dynamics model. This hybrid approach overcomes the inherent limitations of existing techniques, such as the instability of low-pass filters and the inaccuracy of model-based estimations. Furthermore, we integrate a disturbance observer to grant robustness against modeling errors. These two components work in a complementary manner to suppress various error sources. Theoretical analyses and experimental results verify that the proposed method improves robustness and ensures stability. Consequently, the resulting controller achieves more accurate and reliable impedance rendering.