Improvement of CMM throughput using path planning, dynamic lobing error compensation, and structural vibration control.

Abstract

To increase the throughput of a Coordinate Measuring Machine (CMM), while maintaining the same machine accuracy, three issues are addressed: (1) Optimum path planning, (2) touch speed increase, and (3) structural vibration control. The measurement time is reduced by finding the minimum-distance collision-free path. The collision-free path is generated by using the ray-tracing technique in the configuration-space octree database. A new technique is developed to find the optimum path from the silhouette contour of colliding objects using global information. In addition, a fast algorithm is developed to solve for the optimum edge path. The algorithm is implemented on a Sheffield RS-30 CMM with a resulting 50% improvement compared to the maximum-clearance-box method. The dynamic lobing errors associated with the kinematic touch probe need to be reduced in order to increase the touch speed. A compensation model is developed as a function of touch speed and approach angle. The compensation is implemented on a CMM, and test results show a 26% reduced cycle time with higher touch speed. Reduction in the settling time of the time-varying dynamic measurement error was also achieved using structural vibration control. A new adaptive control algorithm is developed by deriving the control law directly from an adaptive lattice filter. The algorithm performs the on-line identification of both the model order as well as the system parameters to compute the input control. Compared to other similar lattice algorithms, the developed algorithm is computationally efficient with a linear computational cost O(N). The algorithm is implemented on a floating-point Digital Signal Processor (the TMS320C30). The proposed algorithm is modified to account for multistep system delays, mismatched input/output signal types, and oscillating input control. These problems are solved by using d-step-ahead predictive control, PID smoothing, and finite differencing the input. The experimental results on a CMM show a 95% reduction in the settling time, and a 91% reduction in the peak-to-peak vibration amplitudes in a time-varying structure.