A dynamic sensing-and modeling approach to three-dimensional point- and area-sensor integration

The recent advancement of 3D non-contact laser scanners enables fast measurement of parts by generating a huge amount of coordinate data for a large surface area in a short time. In contrast, traditional tactile probes in the coordinate measurement machines can generate more accurate coordinate data points at a much slower pace. Therefore, the combination of laser scanners and touch probes can potentially lead to more accurate, faster, and denser measurements. In this paper, we develop a dynamic sensing-and-modeling approach for integrating a tactile point sensor and an area laser scanner to improve the measurement speed and quality. A part is first laser scanned to capture its overall shape. It is then probed via a tactile sensor where the probing positions are dynamically determined to reduce the measurement uncertainty based on a novel next-best-point formulation. Technically, we use the Kalman filter to fuse laser-scanned point cloud and tactile points and to incrementally update the surface model based on the dynamically probed points. We solve the next-best-point problem by transforming the B-spline surface's uncertainty distribution into a higher dimensional uncertainty surface so that the convex hull property of the B-spline surface can be utilized to dramatically reduce the search speed and to guarantee the optimality of the resulting point. Three examples in this paper demonstrate that the dynamic sensing-and-modeling effectively integrates the area laser scanner and the point touch probe and leads to a significant amount of measurement time saving (at least several times faster in all three cases). This dynamic approach's further benefits include reducing surface uncertainty due to the maximum uncertainty control through the next-best-point sensing and improving surface accuracy in surface reconstruction through the use of Kalman filter to account various sensor noise.