To implement robot local calibration techniques, automatic TCF calibration methods for orthogonal tools and sensors are developed, and TCF calibration methods for non-orthogonal end-effectors are introduced. It has been shown that the automatic TCF calibration procedures only take two to three minutes to accurately determine each tool or sensor TCF relative to the robot last joint frame, which greatly reduces the cell initialization time, and increases robot flexibility. Applying only positional information obtained by many combinations of measurement devices, rigid body correction methods have been developed for applications in both two-dimensional and three-dimensional spaces.;Generic data model fitting tools and automatic feature selection techniques are developed. The data model fitting tools can be employed to generate accurate robot performance models for applications where sufficient robot inaccuracy data can be obtained easily and quickly in the local regions of interest. Automatic feature selection techniques use the concept of local regions, robot performance in the region, and designed task tolerances to establish the relationship between the working configurations and the mapping configurations. The selected mapping configurations can then be used to create the calibration task automatically. Automatic feature selection procedures have been successfully implemented in a SEIKO RT-3000 robotic workcell. It has been demonstrated that the automatically generated calibration task together with a linear compensation model improves the robot positioning accuracy to its repeatability.
展开▼
