Pose estimation and material removal study of a non-rigid endoscopic machining tool

Abstract

There is currently no efficient way to quantitatively (precisely) remove material from work pieces in hard-to-reach cavities, such as turbine blades in a jet engine. There is a need to be able to smooth the cracks from the turbine blades with a controlled amount of material removal. Endoscopy is useful for navigating and inspecting features in internal cavities with a flexible body and a semi-rigid bending section. By turning control knobs on the endoscope, the bending section can access a workpiece on its lateral direction. In this research, it is hypothesized that an endoscope can be used to remotely grind a predetermined amount of material and smooth the cracks on a work piece. The research objectives are to set up a precise flexible machining operation and to measure the accuracy of the estimated amount of material that can be removed. Experiment results shows that a reasonable accuracy was achieved using a modified flexible medical endoscope. In this study, two models were built to estimate the pose (position and orientation) and force on a modified PENTAX ES-3801 endoscope. The pose model was based on geometric analysis of the endoscope bending section. When the internal force was compensated in calculation, the average pose estimation error was less than 4 mm of distance and 5 degrees of bending angle. The force model was based on equilibrium analysis of the bending section and the average force prediction error was 0.5 N in the range of 1 to 8 N. The experimental and estimated results prove that the pose and force of the flexible bending section can be predicted and monitored when subjected to varying loads from work piece. In grinding processes, self-excited vibration (chatter) will seriously degrade the geometric accuracy and surface finish of the workpiece. The proposed endoscopic machining tool is naturally vulnerable to chatter due to its high degrees of freedom and low structural rigidity. A dynamic model was built to study the self-excited vibration characteristics of the endoscopic machining tool. By analyzing the stability lobe diagram (SLD), optimized machining parameters were found to reduce the chatter. An on-line chatter detecting method is proposed to identify self-excited vibration in its early stage. Finally, a material removal rate (MRR) model was formulated to predict the amount of material removal in this endoscopic grinding process. By calculating the energy wasted by chatter and the effective energy spent on grinding, the MRR was predicted. Several machining experiments were conducted to validate the repeatability of the model and evaluate its accuracy. The average MRR prediction error is 22% related to the measured MRR. Experiment results reveals that the current MRR prediction accuracy (in the scale of 0.5 mg) is not high enough to meet industry requirement, which is believed to be 0.004 mg for submillimeter size cracks. Recommendations of improving pose estimation and MRR prediction accuracies such as designing a more predictable endoscopic tool structure, using a sphere grinding tool head and adopting a pneumatic driving mechanism or tip mounted motor are proposed.