Adaptive robust control design and experimental demonstration for flexible joint manipulators

A control design for flexible joint manipulator in the presence of nonlinearity and mismatched uncertainty is introduced. The control does not need the possible bound of uncertainty a priori. Only the existence of the bound is assumed. A state transformation is introduced via implanted control to tackle a mismatched system. The scheme utilizes the bounding function by combining states and parameters, which is to be estimated. Then an appropriate parameters update laws are designed to guarantee an asymptotic convergency by adopting Lyapunov approach. The control version shows that states converge to zero for the transformed system, and guarantees the uniform stability and boundness. This is also true for the original system in case either the gravitational force is absent or the system is coordinated such that the gravitational force converges to zero as link angles approach zero. The control performance is verified through experiments and shows an enhanced tracking performance for given references.