An approach to discrete inverse dynamics control of flexible-joint robots

An inverse-dynamics-based method for discrete-time control of flexible-joint robots is presented. The main drawbacks of continuous-time analysis are its computational burden and the necessity for very high sampling frequencies. These inconveniences can be avoided by the use of numerical methods conceived for the solution of systems of differential-algebraic equations. Such an approach naturally leads to a predictive control scheme. As a consequence, in the control process a basic dynamic model of the flexible-joint robot can be used, which is much less complex than in the classical inverse dynamics solution. At the same time, the simplified inverse dynamics approach discussed here accepts low sampling frequencies. Control algorithms are presented and their properties are discussed. Successful experiments on computer control of a two-link manipulator with one flexible joint are described