Abrasive Identification Using Elastomeric Sensing

Kimo Johnson - Chief Science Officer, GelSight, Inc

A grinding and finishing process often involves the careful selection of abrasives and process parameters to achieve a precise surface finish. The surface finish itself might be specified by functional requirements, such as roughness, or to match subjective requirements, such as appearance. Many critical properties of the abrasive, the surface being finished, and the finishing process are revealed by analyzing the microscale topography of the surfaces involved in the process. We will discuss a fast, handheld surface measurement system capable of capturing micron-to-millimeter surface topography of any material, regardless of its optical properties. This system, when combined with modern artificial intelligence methods, can be used for various applications relevant to grinding and finishing processes, including abrasive identification, abrasive lifetime prediction, and surface appearance characterization. This system can also be integrated into a finishing process to provide dynamic assessments of the abrasive, the surface or both.  We will present experimental results on abrasive grit and type identification and abrasive lifetime prediction. People often rely on tactile perception rather than vision for assessing abrasive quality and surface finish. At its core, GelSight technology is a tactile sensor that uses an elastomeric material, imaging sensor, and computer vision algorithms to capture high-resolution 3D data from any surface. The elastomeric material has a thin reflective coating that closely conforms to the topography of the surface on contact. This coating lends its controlled optical properties to the surface allowing for fast and precise 3D measurements at a micron-scale. The resulting data is significantly richer than a profilometer trace and can capture complex surface characteristics. In addition, the rich surface data is directly compatible with modern machine learning methods that were designed to work on images. Despite advancements in manufacturing processes, many aspects of surface finishing still rely on visual and tactile inspection, which is qualitative and often not repeatable across different operators. The properties of abrasives and their effect on surfaces can be captured by micron-scale surface imaging combined with modern artificial intelligence methods. This technology has the potential to digitize and automate a variety of tasks within grinding and finishing processes. 

Back