The non-destructive testing methods offer great benefit in detecting and classifying the weld defects. Among these, infrared (IR) thermography stands out in the inspection, characterization, and analysis of the defects from the camera image sequences, particularly with the recent advent of deep learning. However, in IR, the defect classification becomes a cumbersome task because of the exposure to the inconsistent and unbalanced heat source, which requires additional supervision. In light of this, authors present a fully automated system capable of detecting defective welds according to the electrical resistance properties in the inline mode. The welding process is captured by an IR camera that generates a video sequence. A set of features extracted by such video feeds supervised machine learning and deep learning algorithms in order to build an industrial diagnostic framework for weld defect detection. The experimental study validates the aptitude of a customized convolutional neural network architecture to classify the malfunctioning weld joints with mean accuracy of 99% and median f1 score of 73% across five-fold cross validation on our locally acquired real world dataset. The outcome encourages the integration of thermographic-based quality control frameworks in all applications where fast and accurate recognition and safety assurance are crucial industrial requirements across the production line.

In the last 10 years, the demand for robot-based depalletization systems has constantly increased due to the growth of sectors such as logistics, storage, and supply chains. Since the scenarios are becoming more and more unstructured, characterized by unknown pallet layouts and stock-keeping unit shapes, the classical depalletization systems based on the knowledge of predefined positions within the pallet frame are going to be substituted by innovative and robust solutions based on 2D/3D vision and Deep Learning (DL) methods. In particular, the Convolutional Neural Networks (CNNs) are deep networks that have proven to be effective in processing 2D/3D images, for example in the automatic object detection task, and robust to the possible variability among the data. However, deep neural networks need a big amount of data to be trained. In this context, whenever deep networks are involved in object detection for supporting depalletization systems, the dataset collection represents one of the main bottlenecks during the commissioning phase. The present work aims at comparing different training strategies to customize an object detection model aiming at minimizing the number of images required for model fitting, while ensuring reliable and robust performances. Different approaches based on a CNN for object detection are proposed, evaluated, and compared in terms of the F1-score. The study was conducted considering different starting conditions in terms of the neural network’s weights, the datasets, and the training set sizes. The proposed approaches were evaluated on the detection of different kinds of paper boxes placed on an industrial pallet. The outcome of the work validates that the best strategy is based on fine-tuning of a CNN-based model already trained on the detection of paper boxes, with a median F1-score greater than $85.0\%$.

Smart industrial workstations for the training and evaluation of workers are an innovative approach to face the problems of manufacturing quality assessment and fast training. However, such products do not implement algorithms that are able to accurately track the pose of a hand tool that might also be partially occluded by the operator’s hands. In the best case, the already proposed systems roughly track the position of the operator’s hand center assuming that a certain task has been performed if the hand center position is close enough to a specified area. The problem of the pose estimation of 3D objects, including the hand tool, is an open and debated problem. The methods that lead to high accuracies are time consuming and require a 3D model of the object to detect, which is why they cannot be adopted for a real-time training system. The rise in deep learning has stimulated the search for better-performing vision-based solutions. Nevertheless, the problem of hand tool pose estimation for assembly and training procedures appears to not have been extensively investigated. In this study, four different vision-based methods based, respectively, on ArUco markers, OpenPose, Azure Kinect Body Tracking and the YOLO network have been proposed in order to estimate the position of a specific point of interest of the tool that has to be tracked in real-time during an assembly or maintenance procedure. The proposed approaches have been tested on a real scenario with four users handling a power drill simulating three different conditions during an assembly procedure. The performance of the methods has been evaluated and compared with the HTC Vive tracking system as a benchmark. Then, the advantages and drawbacks in terms of the accuracy and invasiveness of the method have been discussed. The authors can state that OpenPose is the most robust proposal arising from the study. The authors will investigate the OpenPose performance in more depth in further studies. The framework appears to be very interesting regarding its integration into a smart workstation for quality assessment and training.

In automotive industry, and generally in the mass production world, maintenance is a very critical issue requiring special attention since every single stop causes a huge loss in term of item produced due to very small cycle time. Basing on this observations, in the last years, a lot of efforts has been put in failure prevention and condition based maintenance; as an example in Fiat Chrysler Automobiles (FCA) the WCM (World Class Manufacturing) became part of its culture and the area dedicated to Professional Maintenance makes possible many step forwards. The ways WCM reaches the zero breakdown are Time Based Maintenance (TBM) and Condition Based Maintenance (CBM) but further improvements can be reached with focus on cost reduction and by optimizing the component usage without arriving to a fault. In this paper, after an overview of maintenance techniques adopted in FCA plants worldwide, a model-based approach is suggested for a COMAU hemming tool named RHEvo. After the development of a simplified model, we try to estimate the actual status of internal components making use of Neural Network. Focusing on the internal springs, the aging affects the elastic coefficient because of fatigue phenomena. As will be shown, under certain assumptions the cracks presence affects the nominal elastic coefficient; therefore, starting from the estimation coming from the Neural Network, it is possible to model an equivalent crack length. Finally, basing on stochastic crack growth model proposed by Yang and Manning an estimation of internal spring’s Remaining Useful Life Estimation (RULE) is calculated.