Due to inaccuracies in component fabrication and assembly, machine tools and industrial robots have geometric errors (i.e., difference between nominal and actual kinematic motions), which greatly contribute to the errors in parts fabricated on these machines, as well as unduly long process certification times. To compensate for machine tool geometric errors, the standard practice is to directly measure each error individually and, from these measurements, directly populate compensation tables found in the machine tool controller. The drawback to this method is that it is extremely slow due to long instrument set up times and does not capture the complexity (e.g., sagging, twisting) of large machine tools. To compensate for industrial robot geometric errors, circle point analysis is used where the errors of each joint are measured independently. While this method is fast, it still does not capture the complexity of robot kinematic errors. In addition, machine tools and industrial robots suffer from thermal deformations due to changes in ambient temperature and heat sources on the machine, and deflections between the tool and part due to processing forces. These error sources are very difficult to model and, thus, are typically ignored.
This talk will discuss recent work on the volumetric error compensation of large machine tools and industrial robots used for manufacturing tasks A laser tracker is used to measure the machine tool and robot geometric errors over the entire visible joint space. A 6 Degree of Freedom geometric error model is constructed for every joint. Translational and rotational errors for each joint are described by a set of joint-position dependent basis functions and probability-based estimators are employed to identify the geometric error model coefficients. Based on this model, an optimization algorithm is used to populate compensation tables for machine tools, or the inverse Jacobian method is used to modify the joint commands for robots. In this talk we will discuss the details of the new volumetric error compensation methodology and provide several examples of machine tools and robots we have modeled and compensated for a variety of industrial partners. Also, we will discuss our most recent work in on-line compensation of industrial robots where errors are directly measured and compensated for during the operation.
