Integrated tactile-optical coordinate measurement for the reverse engineering of complex geometry

Complex design specifications and tighter tolerances are increasingly required in modern engineering applications, either for functional or aesthetic demands. Multiple sensors are therefore exploited to achieve both holistic measurement information and improved reliability or reduced uncertainty of measurement data. Multi-sensor integration systems can combine data from several information sources (sensors) into a common representational format in order that the measurement evaluation can benefit from all available sensor information and data. This means a multi-sensor system is able to provide more efficient solutions and better performances than a single sensor based system. This thesis develops a compensation approach for reverse engineering applications based on the hybrid tactile-optical multi-sensor system.

In the multi-sensor integration system, each individual sensor should be configured to its optimum for satisfactory measurement results. All the data measured from different equipment have to be precisely integrated into a common coordinate system. To solve this problem, this thesis proposes an accurate and flexible method to unify the coordinates of optical and tactile sensors for reverse engineering. A sphere-plate artefact with nine spheres is created and a set of routines are developed for data integration of a multi-sensor system. Experimental results prove that this novel centroid approach is more accurate than the traditional method. Thus, data sampled by different measuring devices, irrespective of their location can be accurately unified.

This thesis describes a competitive integration for reverse engineering applications where the point cloud data scanned by the fast optical sensor is compensated and corrected by the slower, but more accurate tactile probe measurement to improve its overall accuracy. A new competitive approach for rapid and accurate reverse engineering of geometric features from multi-sensor systems based on a geometric algebra approach is proposed and a set of programs based on the MATLAB platform has been generated for the verification of the proposed method. After data fusion, the measurement efficiency is improved 90% in comparison to the tactile method and the accuracy of the reconstructed geometric model is improved from 45 micrometres to 7 micrometres in comparison to the optical method, which are validated by case study.