Generalized learning-based fuzzy environment model

Simulations on impedance model following force control using generalized learning-based fuzzy environment model is presented for articulated industrial robots with an open architecture controller. The desired damping, which is one of the impedance parameters, mainly has an effect on force control performance. An important goal for successfully using the conventional impedance model following force control method is how to suitably tune the desired damping according to each environment and task. However, there are no systematic tuning methods, so that the desired damping is generally tuned by trial and error. To overcome this problem we have already introduced an approach that produces the desired time-varying damping, giving the critical damping condition in contact with an object. However, considering the critical damping condition requires that the physical parameters of the object, such as viscosity and stiffness, are beforehand known. In this chapter, generalized fuzzy environment model (GFEM) is proposed to deal with unknown environments. The GFEM is designed by integrating several FEMs. Each FEM is optimized by using genetic algorithms under several known environments. The GFEM not only estimates the stiffness of unknown environment but also systematically outputs the desired time-varying damping for stable force control. The time-varying desired damping allows the impedance model following force controller to have an ability to suppress overshoots and oscillations in force control. The effectiveness and promise are demonstrated through hybrid position/force control simulations using the dynamic model of PUMA560 manipulator.