Displacement Control of Water Hydraulic McKibben Muscles with Load Compensation

Abstract

This paper is concerned with displacement control of water hydraulic McKibben muscle. In general, control performance of conventional PI control is not enough because the muscle has strong nonlinearities. Moreover, a load connected with the muscle affects in the performance. For these problems, we propose combination of model-based control and adaptive parameter estimation. First, model predictive control is introduced and its control performance is validated. However, the performance is degraded under loaded condition because the characteristics of the muscle are changed and the one-step-ahead estimation, which is used to generate control inputs, should fail. In order to compensate the effect of characteristic change under the condition, we combine an adaptive parameter estimation algorithm with the proposed control. In particular, recursive least squares algorithm is applied as an adaptive algorithm. As a result, effect of load can be compensated and the performance under loaded condition can be improved.