Adaptive lattice estimation and control of a manipulator with one flexible forearm

An application of recursive covariance lattice algorithms to the adaptive estimation and control of a manipulator with one flexible link is presented. These algorithms are a set of pure order recursive lattice equations, which can in principle identify the effective order and the corresponding parameters of an ARMA prediction model of the manipulator. The reduced-order prediction model represents the significant dynamics of the plant, and is used to generate control sequences by minimising a weighted performance index. In the simulations, the manipulator is modelled by the finite element method and Langrange´s equations. The performance and robustness of the lattice filter and the adaptive controller is demonstrated by numerical results.