A feedforward decoupling concept for the control of elastic robots

A reference-oriented procedure for controlling elastic robots is presented. This affords, in a first step, an optimal path planning method generating the reference trajectory. It will be realized by a classical feedforward decoupling scheme. The elastic deviations from this reference are corrected within the feedforward loop by calculation of the joint correction angles putting back the endpoint to its nominal position. In a final step, elastic vibrations of the arms are damped by an additional joint output control system, which uses as an input joint kinematics and strain gauge measurements of the elastic curvatures. The concept has been realized for a laboratory manipulator. Theory and experiments agree very well.