Ebook: Design Studies and Intelligence Engineering

The technologies applied in design studies vary from basic theories to more application-based systems, and intelligence engineering techniques, such as computer-aided industrial design, human factor design, and greenhouse design, also play a significant role in design science. Intelligence engineering technologies encompass both theoretical and application perspectives such as computational technologies, sensing technologies, and video detection. Intelligence engineering is multidisciplinary in nature, and promotes cooperation, exchange and discussion between organizations and researchers from these diverse fields.

For this collection of papers on design studies and intelligence engineering, 210 submissions were reviewed. The decision of the conference chairs was based on the scores given to each paper by three or four referees and 50 papers were accepted for publication in this volume. The 2021 International Symposium on Design Studies and Intelligence Engineering (DSIE2021) was held in Hangzhou, China on November 27–28, 2021. The annual conference invites renowned experts from around the world to speak on their specialist topics, providing a platform for many professionals and researchers from industry and academia to exchange and discuss recent advances in the field of design studies and intelligence engineering. DSIE 2021 is sponsored by Zhejiang Sci-Tech University and co-sponsored by the Chinese Mechanical Engineering Society-Industrial Design, the Aurel Vlaicu University of Arad, Techno India College of Technology, and Tanta University.

We thank all contributors for their efforts in submitting their manuscripts on time. We would also like to express our gratitude to the reviewers for their support, and gratefully acknowledge their contribution to the timely completion of this volume.

Lakhmi C. Jain

Valentina Emilia Balas

Qun Wu

Fuqian Shi

Music emotion recognition (MER) studies have made great progress in detecting the emotions of music segments and analyzing the emotional dynamics of songs. The overall emotion and depth information of entire songs may be more suitable for real-life applications in certain scenarios. This study focuses on recognizing the overall emotion and depth of entire songs. First, we constructed a public dataset containing 3839 popular songs in China (PSIC3839) by conducting an online experiment to collect the arousal, valence, and depth annotation of each song. Second, we used handcrafted feature-based method to predict the overall emotion and depth values. Support vector regressions using Mel frequency cepstrum coefficients features as inputs achieve good model performance (arousal: R² = 0.609; valence: R² = 0.354; and depth: R² = 0.465). Finally, the groupwise and personalized results were also investigated by training a unique regressor for each group or individual, which provides a reference for future research.

With the in-depth application of semiotics in product design, it has brought great changes to modern design process. Starting from the noise nodes generated in the product design process, taking the actual case of a small rice mill as an example, starting from the aspects of structure, semantics and pragmatics, through the combination of theory and practice. Finally, the feasibility and practical value of noise control in symbol transmission channel are verified.

Intelligent manufacturing for the fabric dyeing industry requires high-performance dyeing recipe recommendation systems. Nowadays, recommending dyeing recipes by mining dyeing manufacturing data has become a new direction for the development of recipe recommendation systems. As one of the indispensable parts in the system development, data pre-processing needs more than routine steps such as the removal of missing data and outliers. Considering that dyes can have very different coloration properties on different fabrics, dyeing manufacturing records for a given dye combination to different fabric types should be properly categorized before they are used for training regression models for dyeing recipe prediction. In this paper, we propose a simple but effective method for this categorization work. Our method uses conventional K-means clustering analysis to find fabric types that have similar coloration properties for a given dye combination. We have applied the method on a dye combination formed by Colvaceton reactive dye-navy blue CF (CRD-navy blue), Colvaceton reactive dye-bright red 3BSN150% (CRD-red) and Colvaceton reactive dye-yellow 3RS150% (CRD-yellow) on 28 different types of fabrics. We show that these 28 types of fabrics can be well categorized into 8 groups based on the coloration properties. Our proposed method can be listed as one of the standard data pre-processing steps in the development of data-mining based recipe recommendation systems.

This paper proposes an algorithm to improve the efficiency of the multi-robot system in simulated global information map containing obstacles. The designed multi-AGV scheduling algorithm is based on an optimal shortest path algorithm with the combination of the waiting mode and motion coordination. The proposed shortest path algorithm not only has lower time delay but also decreases the possibility of collision of the multi-robot system. In addition, simulated global information maps are established to test the efficacy of the algorithm.

Aspect sentiment triplet extraction (ASTE) is a relative difficult and novel research, which is a subtask of aspect-based sentiment analysis (ABSA). ASTE is a task that extracts triplets of aspects being discussed, relevant opinion entities and sentiment polarities from a given sentence. Existing approaches mainly deal with this problem by pipeline or simple multi-task structure, which do not take full advantage of the strong correlation among the three elements of the triplet. In this work, we adopt two special tagging schemes, AOBIO and Pair Tagging Scheme (PTS), and propose an efficient end-to-end multi-task model named Joint Aspect Sentiment Triplet Extraction (JASTE) to address ASTE task. JASTE is composed of three modules: aspect and opinion extraction module, relation module and sentiment module. Specially, the relation module is designed to capture the relationship between aspect and opinion properly. The three modules interact with each other by sharing the same embedding. Extensive experimental results on different benchmark datasets show that JASTE can significantly outperform state-of-the-art performances.

As one of the world’s few large-scale aquaculture countries, the mechanization, output and efficiency are low in Chinese aquaculture industry.with the development of industrial automation technology and computer image processing technology, aquaculture will be automated, scientific management transformation. As a specific application of computer recognition technology, underwater image recognition technology is important to promote aquaculture industry automation and intelligence. In this paper, first of all introduces the existing research of underwater target image recognition, mainly presents the underwater image recognition technology based on deep learning. And then the current problems of underwater image recognition is summarized. Besides, the future development of underwater image recognition technology and its application in wisdom fishery is also discussed, which provides other scholar with some theoretical help and a comprehensive reference.

Aiming at the problems of low efficiency, low accuracy and long maintenance cycle of traditional manual detection methods for switch gear detection, robot is proposed to intelligently detect. Through the analysis on sizes of switch gear and switch room in electric power system, switch gear live detection robot is designed to automatically complete switch gear’s real time live detection. While using D-H parameter method to describe mechanical arm, switch gear live detection robot’s simulations are carried out by MATLAB, which analyse the spatial location of mechanical arm. Designed detection robot can ensure high efficiency and reduce manpower input.

The development of intelligent industrialization has gradually increased the demand for industrial robots, and it has also promoted the emergence and development of industrial robot protective clothing. At this stage, the design methods of sewing patterns are limited to planes and lack simulation stitching and try-on in three-dimensional space. In this study, the virtual try-on of protective clothing for industrial robots was realized in the CLO3D environment, and the fabric was simulated using tools in the software. First, import the digital model of the industrial robot into CLO3D, and copy the AUTO CAD template in CLO3D. Then, input the results of the fabric test to simulate the fabric. Use the software’s virtual stitching and virtual wearing to observe the wearing effect of protective clothing and the data-based fabric pressure, directly modify the model and observe the modification effect. After research, the three-dimensional simulated fitting can clearly show the characteristics of looseness or tightness and whether it is convenient to put on and take off. The model can be adjusted appropriately in time, which can reduce costs and improve production efficiency. It is an important exploration for the development of industrial robot protective clothing.

In order to explore a smart bracelet design strategy that is more suitable for the elderly, this paper studies relevant literatures and related products of smart bracelet at home and abroad, and sorts out basic concepts and related technologies of implicit interaction theory. Through questionnaire survey and semi-structured interview, we summarized. the basic characteristics of the elderly and the corresponding pain points of using smart wristband. We built a smart bracelet design scheme for the elderly based on implicit interaction theory, which provides certain theoretical support and practical reference value for the follow-up study on the age- suitability design of smart bracelet.

Global warming has attracted more and more people’s attention. Since products are one of the main sources of GHG emission, the firm is seeking appropriate methods to reduce GHG emission of the product. At present, product family design is widely adopted for meeting the various demand of customers. To reduce the GHG emission of products, some methods have been proposed for low-carbon product family design in recently years. In existing research, the related data of low-carbon product family design is given as crisp value. However, in a real environment, some design data can’t be assessed accurately. To this end, this paper proposes a uncertain optimization model for low-carbon product family design. In the model, the related uncertain data for low-carbon product family design is given as interval numbers. Based on the objective of profit and GHG emission, the model can simultaneously determine product family configuration, supplier selection and price strategies of product variants. In addition, the genetic algorithm is developed to solve the established model. Finally, a case study is performed to demonstrate the effectiveness of the proposed approach.

Guiding by Activity Theory and previous studies, this study designed a Social Components Framework of Educational Game (SCFEG) as the theory-based mechanism or framework to guide the systematical design of social attributes for the Traditional Chinese Medicine (TCM) popular science game. By adopting SCFEG, this study depicts the initial design pictures of social structure of the first and second outposts of “Huaxia Medicine Fairy”. This study may bring some guidance and directions for the designers of popular science games, educational games or serious games.

In the context of defect detection, there is a difficulty in collecting annotated datasets, which leads to limited labeled data. In addition to this, most of the defect detection methods have the problem of missing detailed information about the defects. To cope with these problems, this paper shows a dense differential Siamese network structure for the defect detection of stamping manufacture. In the stamping setting, the foreground of the image frequently changes, while the background remains the same. Based on this finding, we separate the encoding layer of the network into two streams with the same structure and shared weights, so that we can handle the foreground and background image pairs simultaneously. To extract detailed information of defects, we also impose the dense skip connections into our network. Through these skip connections, we can obtain different levels of semantic information and capture more detailed information about the defects. Testing results on the defect dataset collected from real stamping machines show that our method significantly improves over other state-of-the-art methods on several evaluation metrics.

Deep neural networks have recently been used to address surface anomaly detection in industrial quality control and have achieved much success. However, addressing the data imbalance problem, especially the Easy/Hard Examples (EHE) imbalance problem, remain a challenging task in anomaly detection. To alleviate this problem, we propose a two-stage convolutional neural network with a novel loss function, i.e., concentrated loss function. Specifically, the concentrated loss function enables the model to pay more attention to hard examples and improve the quality of segmentation for imbalanced data. To verify the effect of our method, we implement our method on the surface anomaly detection dataset, i.e., the KolektorSDD2 dataset. The experimental results show the superiority of our method over the other state-of-the-art approaches.

Filters pruning methods are widely used to accelerate the inference process of Deep Neural Networks (DNNs). Among them, Soft Filter Pruning (SFP) has achieved increasing attention due to its compatibility and flexibility. However, conventional SFP directly sets the pruned filters to zero during training phase, discarding the training information of the pruned filter completely. In order to solve the above problem, this paper proposed a soft and smooth pruning method to retain the training information among pruned filters. Instead of the zeroing the pruned filters, our method imposes a filter weakening strategy, which gradually forces the pruned filters to zeros. Such a gradual pruning framework will give the pruned model more chances to recover the lost information and boost the pruning performance. To verify the the actual effectiveness of our proposed method, we conduct several experimental results on one dataset of metal surface images captured in a controlled industrial environment, i.e.,KSDD, using Resnet-20 with various pruning rates. Experimental results show that our filter weakening strategy consistently achieves superior performance over the compared methods, especially when a large amount of filters is pruned. By pruning 60% of filters, our method only drops 2.52% on Accuracy.

This study constructs a new mixed reality MR holographic teaching application course, summarizes its more than 5 years’ research and practical experience in the digital media direction of Wenzhou Business School, and discusses the universality and shortcomings of mixed reality MR holographic teaching to digital media majors and other domestic university digital media majors combined with the needs of the society and the characteristics of the digital media direction of visual communication design major. The practical exploration of this teaching model and method not only improves the construction effectiveness and talent training quality of the mixed reality MR holographic teaching course in the digital media major direction of Wenzhou Business School, but also plays a leading and exemplary role in the construction and demonstration of related professional courses in other colleges and universities.

With the development of machine learning, artificial intelligence and other fields, the processing of data mining has become more and more complex. As a data preprocessing step, feature selection is very important in many tasks, like classification, clustering and regression, etc. However, traditional feature selection methods learns similarity matrix from original data to calculate relevant data. What this method learns is the relationships which are linear between data and their labels, and it cannot deal with complex nonlinear data well in real-world applications. In this article, we proposed a feature selection method based on neural network that can select discriminative feature subsets by neural network pruning. And update all weights by gradient descent. Experimental results of our method on several real-world datasets achieve competitive or superior performance compared to three close related feature selection approaches.

In this article, based on the self-represented multi-view subspace clustering framework, we propose a new clustering model. Based on the assumption that different features can be linearly represented by data mapped to different subspaces, multiview subspace learning methods take advantage of the complementary and consensus informations between various kind of views of the data can boost the clustering performance. We search for the tensor with the lowest rank and then extract the frontal slice of it to establish a well-structured affinity matrix. Based on the tensor singular value decomposition (t-SVD), our low-rank constraint can be achieved. We impose the ℓ_2,p-norm to flexibly control the sparsity of the error matrix, making it more robust to noise, which will enhance the robustness of our clustering model. With combining ℓ_2,p-norm and tensor multi-rank minimization, the proposed Multi-view Subspace Clustering(MVSC) model can effectively perform clustering with multiple data resources. We test our model on one real-world spoon dataset and several publicly availabe datasets. Extensive evaluation methods have proved that our model is effective and efficient.

At present, there may be some problems in the production process of spoon, such as the lack of material on the surface of spoon. In order to effectively detect the surface defects of spoon, a defect detection method based on improved YOLO V3 model is proposed in this paper. Firstly, the output layers of the second and third residual blocks in the backbone network Darknet-53 are selected to build the feature pyramid network, which shortens the transmission path of feature information. In this case, we can better retain the feature information of small target defects. Secondly, the anchor boxes is adjusted to strengthen the ability of the model for small target defects detection. We test the proposed method on one spoon defect dataset, which is collected from the real-world industry manufactory scenario. The results show that the average precision of our algorithm reaches at 95.14%, which is better than the conventional YOLO V3 algorithm by 9.35%. Meanwhile, our algorithm is 9.12% faster than YOLO V3 with a 32.3 fps detection speed, demonstrating its efficiency and effectiveness for spoon defect detection.

In many practical applications, data are represented by high-dimensional features. Although the traditional K-means algorithm is simple, it usually gets the approximation solution by eigenvalue decomposition, this method led to the model less efficient. In addition, their loss functions are sensitive to data distribution. In this paper, a clustering model of adaptive K-means with sparse constraints is proposed. The proposed method is designed by combining the dimension reduction with sparse constraints and adaptive clustering. It provides a flexible computational framework for subspace clustering and is suitable for different distribution data sets. Besides, the sparsely constraint in our method can remove redundant features and retain useful information. We develop an effective alternative optimization algorithm to solve our model. Finally, the extended experiments on several benchmark datasets demonstrate the advantages of our method over other clustering algorithms

A method to design the insoles for pressure relief based on the plantar pressure distribution characteristics by FEM (Finite Element Method), so as to realize the customized insole service for different individuals. By establishing a composite model of foot bones and soft tissues, human foot simulation was conducted to obtain the plantar pressure distribution data of latex, sponge, TPU and EVA flat insoles respectively; MATLAB was used to binarize the plantar pressure distribution map obtained from the simulation and extract the contour of the plantar pressure concentration area for the design of pressure relieving insoles, and the design of insoles for pressure relieving insoles was carried out according to the different properties of the four materials. The design of the insole combination is based on the different properties of the four materials. The appropriate solution of Insole is proposed by this design process as a prototype so as to achieve an intelligent approach for footwear design.

In order to inherit and innovate traditional weft-backed structure jacquard fabric, a design principle and method of digital jacquard with 2:1 weft-backed structure was proposed. Based on the in-depth analysis of the characteristics of the traditional 2:1 weft-backed structure, four digital structural models (A1, A2, B1 and B2) with shaded weave were built by using digital jacquard technology. It lays a theoretical foundation for the creation of four kinds of digital jacquard fabrics with colour shading effects. In addition, the combined application of model A1 and model B1 was taken as an example to illustrate the design method. The resulting fabric had a delicate and natural colour shading effect. And its main and minor motifs were clearly distinguished. Meanwhile, the colour expression was increased while the yarn utilisation and production efficiency were improved. The principle and method proposed in this paper can create a wide variety of digital jacquard fabrics with colour shading effects, which provides ideas for the structural innovation and variety design of traditional jacquard fabrics.

The design of reasonable reward mechanisms in exercise games aims to increase people’s intrinsic motivation to participate in the games for training so that they can persist for a long time and achieve exercise results. This study developed two game modes based on a prototype of an exercises game for a Kinect interactive device. Two reward mechanisms, in-game rewards and extrinsic rewards, were included in one of the game modes respectively. Twelve older subjects and twelve younger subjects were recruited to test the game. They rated the game using an intrinsic motivation scale. The results showed that young people who played the in-game reward mode scored higher on the motivation scale. In contrast to young people, older people had higher motivation levels when playing the extrinsic reward mode. The results highlight the variability in the effects of in-game and extrinsic rewards on intrinsic motivation across age groups and can help people better design reward mechanisms when developing exercise games.

In order to explore the effect of different textures on human emotions under different sensory conditions, we collected the effective tendency of textures of six types from 18 to 30 years old subjects in visual, tactile and auditory experiments through psycho-physical experiment. The results show that the subjects have different emotional feelings for textures under different senses. But in all experiments, wood is a relatively safe and reliable textures in subjects’ cognition. This essay, through the study of textures, aims to raise design suggestions on the use of large medical equipment.

To help the disabled people eat, this paper proposed a AI meal-assisting manipulator via the detection of the open-mouth target. The YOLO_V3 algorithm was used to complete the construction of the mouth opening and closing state data set and model training, and the mouth state recognition model was successfully constructed and evaluated. The experimental results show that the model has a high recognition accuracy rate, accuracy rate and recall rate of the mouth opening and closing state. Finally, the application verification was successfully achieved by building a functional prototype of the meal-assisting manipulator.