Robust Control for Industrial Robots

Summary

The control of industrial robots is a demanding problem as we have strongly non-linear multi-degree-of-freedom systems, disturbances like elasticity or parameters varying within a range of some 100 per cent (e.g. due to payload changes). The object of this paper is to design a control which is insensitive (robust) with respect to parameter variations or disturbances such as non-linear coupling and elasticity. This goal is achieved by following a proposition of LEITMANN, BREINL: a linear control (Riccati-regulator) designed for a nominal system is superposed by a non-linear control compensating the effects of the unknown parameters. The general theory is specified for SCARA-type robots, including such effects as elastic gears (‘harmonic drives’) and elastic arms. The control is completely decentralized, i.e. decoupled for the different axes. This is easy to realize and is therefore well suited for real-time computation. To show the efficiency of the proposed control concept, simulations for an exemplary manoeuvre are presented. If we increase the payload mass for 1000 per cent and reduce the gear elasticity for 10 per cent, a classical control would show intolerable overshoot and motion times. The robust control, however, is able to guarantee no overshoot and only small motion time prolongations.