Non-Linear Control of Robot aided Belt Grinding Manufacturing Processes

In the production process of faucets, large tolerances, mainly caused by the casting process of the raw part, lead to wide variations in faucet geometry. Additional effort is thus required to manually set the parameters for the robot-guided grinding process. As design products with a complex geometry consisting of several convex and concave surface elements, faucet manufacture poses significant challenges for the robot programmer.Our contribution towards solving this problem is an analytic description of the manufacturing process to improve the parameters. The developed model of the grinding process consists of a belt grinding machine with contact wheel and the robot which holds the workpiece during the entire manufacturing process. By means of Perturbation Theory, these systems can be divided into a slow and a fast process. The multibody dynamic system of this model consists of the robot and grinding machine dynamics and additionally of the grinding machine-robotinteraction. This interaction induces a strong nonlinearity, namely the possible loss of contact. Moreover, the behavior of the robot itself is nonlinear, which results in varying compliances depending on the actual position, orientation and trajectory of the robot. For the theoretical analytical model, these influences are respected. In order to validate the derived system model, an experimental setup was designed and analyzed to parameterize, or estimate, significant parameters within the model. As a consequence of the applied research, three representations of the grinding process were obtained and the fundamentals are provided to improve the process by means of control systems.