THE EFFECT OF DIFFERENT OCCUPATIONAL BACKGROUND NOISES ON VOICE RECOGNITION ACCURACY

Voice recognition has become an integral part of our lives, commonly used in call centers and as part of virtual assistants. However, voice recognition is increasingly applied to more industrial uses. Each of these use cases has unique characteristics that may impact the effectiveness of voice recognition, which could impact industrial productivity, performance, or even safety. One of the most prominent among them is the unique background noises that are dominant in each industry. The existence of different machinery and different work layouts are primary contributors to this. Another important characteristic is the type of communication that is present in these settings. Daily communication often involves longer sentences uttered under relatively silent conditions, whereas communication in industrial settings is often short and conducted in loud conditions. In this study, we demonstrated the importance of taking these two elements into account by comparing the performances of two voice recognition algorithms under several background noise conditions: a regular Convolutional Neural Network (CNN) based voice recognition algorithm to an Auto Speech Recognition (ASR) based model with a denoising module. Our results indicate that there is a significant performance drop between the typical background noise use (white noise) and the rest of the background noises. Also, our custom ASR model with the denoising module outperformed the CNN based model with an overall performance increase between 14-35% across all background noises. . Both results give proof that specialized voice recognition algorithms need to be developed for these environments to reliably deploy them as control mechanisms.

A surface defect detection method via fusing multi-level features

Surface defect detection is very important to ensure product quality, but most of the surface defects of industrial products are characterized by low contrast, big size difference and category similarity, which brings challenges to the automatic detection of defects. To solve these problems, we propose a defect detection method based on convolutional neural network. In this method, a backbone network with semantic supervision is applied to extract the features of different levels. While a multi-level feature fusion module is proposed to fuse adjacent feature maps into high-resolution feature maps successively, which significantly improves the prediction accuracy of the network. Finally, an Encoding module is used to obtain the global context information of the high-resolution feature map, which further improves the pixel classification accuracy. Experiments show that the proposed method is superior to other methods in NEU_SEG (mIoU of 85.27) and MT (mIoU of 77.82) datasets, and has the potential of real-time detection.

Computational Design for Digitally Fabricated 3D Inductive Power Transfer Coils

The geometric shapes and the relative position of coils influence the performance of a three-dimensional (3D) inductive power transfer system. In this paper, we propose a coil design method for specifying the positions and the shapes of a pair of coils to transmit the desired power in 3D. Given region of interests (ROIs) for designing the transmitter and the receiver coils on two surfaces, the transmitter coil is generated around the center of its ROI first. The center of the receiver coil is estimated as a random seed position in the corresponding 3D surface. At this position, we use the heatmap method with electromagnetic constraints to iteratively extend the coil until the desired power can be transferred via the set of coils. In each step, the shape of the extension, i.e. a new turn of the receiver coil, is found as a spiral curve based on the convex hulls of adjacent turns in the 2D projection plane along their normal direction. Then, the optimal position of the receiver coil is found by maximizing the efficiency of the system. In the next step, the position and the shape of the transmitter coil are optimized based on the fixed receiver coil using the same method. This zig-zag optimization process iterates until an optimum is reached. Simulations and experiments with digitally fabricated prototypes were conducted and the effectiveness of the proposed 3D coil design method was verified. Possible future research directions are highlighted well.

User-Requirements Analysis on Augmented Reality-based Maintenance in Manufacturing

Maintenance of technical equipment in manufacturing is inevitable for sustained productivity with minimal downtimes. Elimination of unscheduled interruptions as well as real-time monitoring of equipment health can potentially benefit from adopting augmented reality (AR) technology. How best to employ this technology in maintenance demands a fundamental comprehension of user requirements for production planners. Despite augmented reality applications being developed to assist various manufacturing operations, no previous study has examined how these user requirements in maintenance have been fulfilled and the potential opportunities that exist for further development. Reviews on maintenance have been general on all industrial fields rather than focusing on a specific industry. In this regard, a systematic literature review was performed on previous studies on augmented reality applications in the maintenance of manufacturing entities from 2017 to 2021. Specifically, the review examines how user requirements have been addressed by these studies and identifies gaps for future research. The user requirements are drawn from the challenges encountered during AR-based maintenance in manufacturing following a similar approach to usability engineering methodologies. The needs are identified as ergonomics, communication, situational awareness, intelligence sources, feedback, safety, motivation, and performance assessment. Contributing factors to those needs are cross-tabulated with the requirements and their results presented as trends, prior to drawing insights and providing possible future suggestions for the made observations.

Online Maintenance Prioritization via Monte Carlo Tree Search and Case-based Reasoning

When maintenance resources in a manufacturing system are limited, a challenge arises in determining how to allocate these resources among multiple competing maintenance jobs. We formulate this problem as an online prioritization problem using a Markov decision process (MDP) to model the system behavior and Monte Carlo tree search (MCTS) to seek optimal maintenance actions in various states of the system. Further, we use Case-based Reasoning (CBR) to retain and reuse search experience gathered from MCTS to reduce the computational effort needed over time and to improve decision-making efficiency. We demonstrate that our proposed method results in increased system throughput when compared to existing methods of maintenance prioritization while also reducing the time needed to identify optimal maintenance actions as more experience is gathered. This is especially beneficial in manufacturing settings where maintenance decisions must be made quickly.

Identify Finger Rotation Angles with ArUco Markers and Action Cameras

Measuring the motions of human hand joints is often a challenge due to the high number of degrees of freedom. In this study, we proposed a hand tracking system utilizing action cameras and ArUco markers to continuously measure the rotation angles of hand joints. Three methods were developed to estimate the joint rotation angles. The pos-based method transforms marker positions to a reference coordinate system (RCS) and extracts a hand skeleton to identify the rotation angles. Similarly, the orient-x-based method calculates the rotation angles from the transformed x-orientations of the detected markers in the RCS. In contrast, the orient-mat-based method first identifies the rotation angles in each camera coordinate system using the detected orientations, and then, synthesizes the results regarding each joint. Experiment results indicated that the repeatability errors with one camera regarding different marker sizes were around 2.64 to 27.56 degrees and 0.60 to 2.36 degrees using the marker positions and orientations respectively. When multiple cameras were employed to measure the joint rotation angles, the angles measured by using the three methods were comparable with that measured by a goniometer. Despite larger deviations occurred when using the pos-based method. Further analysis indicated that the results of using the orient-mat-based method can describe more types of joint rotations, and the effectiveness of this method was verified by capturing hand movements of several participants. Thus it is recommended for measuring joint rotation angles in practical setups.

A review of Virtual Assistants' characteristics: recommendations for designing an optimal human-machine cooperation

Designed to improve human-machine interactions, virtual agents, and particularly virtual assistants (VAs), are spreading in our daily lives. Presenting a very wide variety of characteristics, studies generally report their own agent with its own characteristics and objective. So we can wonder if some of these characteristics are a consensus for VAs in general. Within this work, we aim to identify the agents' characteristics that should be considered when designing a virtual assistant promoting the best communication and cooperation between man and machine. We review the aspects of representation of the agent (embodied or not) and its ability to interact with the human being whether by speech or gestures, but also by displaying personality traits. This overview makes some focuses on virtual assistance of any kind embarked on vehicles.

Automated and adaptive geometry preparation for ar/vr-applications

Product visualization in AR/VR applications requires a largely manual process of data preparation. Previous publications focus on error-free triangulation or transformation of product structure data and display attributes for AR/VR applications. This paper focuses on the preparation of the required geometry data. In this context, a significant reduction in effort can be achieved through automation. The steps of geometry preparation are identified and examined concerning their automation potential. In addition, possible couplings of sub-steps are discussed. Based on these explanations, a structure for the geometry preparation process is proposed. With this structured preparation process, it becomes possible to consider the available computing power of the target platform during the geometry preparation. The number of objects to be rendered, the tessellation quality, and the level of detail can be controlled by the automated choice of transformation parameters. Through this approach, tedious preparation tasks and iterative performance optimization can be avoided in the future, which also simplifies the integration of AR/VR applications into product development and use. A software tool is presented in which partial steps of the automatic preparation are already implemented. After an analysis of the product structure of a CAD file, the transformation is executed for each component. Functions implemented so far allow, for example, the selection of assemblies and parts based on filter options, the transformation of geometries in batch mode, the removal of certain details, and the creation of UV maps. Flexibility, transformation quality, and timesavings are described and discussed.

A Comparative Study of Virtual Footwear Try-on Applications in Virtual and Augmented Reality

Virtual try-on technology (VTO) in virtual reality (VR) and augmented reality (AR) has been developed for years to create novel shopping experiences for users by allowing them to virtually wear fashion products. Compared to garments or facial accessories, fewer studies have focused on virtual footwear try-on, regardless of user study or technical development. Thus, it is necessary to examine the effectiveness of existing VTO applications on the user's affective responses. In this study, we compared the user experience of three different footwear try-on methods (real, VR, and AR) with both physiological and psychological measures. Subjects conducted a try-on experiment on different pairs of sneakers. Each subject's gaze trajectory was recorded using an eye tracker and analyzed to show his/her visual attention in each method. Afterward, the subjects completed questionnaires to assess the sense of presence, usability, and the user experience score for the try-on processes, and subsequently attended a think-aloud procedure to express their thoughts. The analysis results of the collected data showed that the user experience produced by the VR and AR try-on is not comparable to that of the real environment. The results also revealed factors that negatively affect the quality of the user's interaction with the processes. These findings may provide insights into further improvements in VTO technology.

A Genetic Algorithm and RNN-LSTM model for Remaining Battery Capacity Prediction

Li-ion batteries have diversified applications in everyday life. The temperature change, overcharging, over-discharging is playing critical roles in affecting battery life in a significant manner. In this paper, the deep learning-based method is applied for the prognostics of a single Li-ion battery. The proposed design uses a recurrent neural network variant, Long short term memory. The model's parameters are optimized through a Genetic Algorithm based parameter selector The method applies to a sequence of data values comprising of the voltage, the charge capacity, the current, and the temperature. The estimation of battery capacity is not only based on the current or defined state of the battery; instead, it is generated on the complete data profile. The robustness of the model is tested by comparing with techniques such as Support vector regressor, Kalman Filter, neural networks on normal and noisy test sets. The paper also proposes a feature selection and engineering scheme for battery capacity prediction. The proposed model outperforms the techniques available in literature with high generalization to noise and other perturbations. The model is independent of the section of charging curve used for prediction of battery capacity. Various experimentation has been conducted on the model and the results have been validated.