Sebti Foufou

Convolutional neural networks (CNNs) are useful tools for analyzing images in several facets of computer vision. Compared to 2D images, 3D points are not represented as a regular grid but rather as a collection of irregularly spaced points in three dimensions, varying significantly depending on the object or scene being described. This makes applying convolutions to 3D points challenging. This work presents a new method called SubConvFusion3D for applying convolutions on a 3D point cloud. By considering the multiple scales of the 3D points, the proposed method minimizes confusion and redundancy while extracting different features at different scales and detail levels. To achieve this, different sub-convolutional networks are first used, allowing for independent analysis. It enables 1D convolutions to extract features for every subnetwork at various scales. The provided features are finally combined to form the final feature vector, by using a new merging (fusion) function. The experimental results conducted on the ModelNet40 dataset show competitive performance achieved by our methodology. The training of SubConvFusion3D achieved an outstanding training accuracy of 87.2%, demonstrating our approach's competitive performance.

Developing a more automated industrial digital thread is vital to realize the smart manufacturing and industry 4.0 vision. The digital thread allows for efficient sharing across product lifecycle stages. Current techniques are not robust in relating downstream data, such as manufacturing and inspection information, back to design for better decision making. We previously presented a methodology that aligns numerical control (NC) code, a standard for representing machine tool instructions, to controller data represented in MTConnect, a standard that provides a vocabulary for generalizing execution logs from different machine tools and devices. This paper extends our previous work by automating the tool identification using a k-means clustering algorithm to refine the alignment of the data. In doing so, we compare different distance techniques to analyze the spatial-temporal registration of the two datasets, i.e., the NC code and MTConnect data. Then, we assess the efficiency of our method through an error measurement technique that expresses the quality of the alignment. Finally, we apply our methodology to a case study that includes verified process plans and real execution data, derived from the smart manufacturing systems test bd hosted at the National Institute of Standards and Technology. Our analysis shows that dynamic time warping achieves the best point registration with the least errors compared with other alignment techniques.