Integrating aesthetic and emotional preferences in social robot design: An affective design approach with Kansei Engineering and Deep Convolutional Generative Adversarial Network

Recently, many companies have increasingly emphasized product appearance aesthetics and emotional preference-based design to enhance the competitiveness and popularity of their products. Identifying the interaction between product appearance and customer preferences and mining design information from the interacting context play essential roles in affect-related design approaches. However, due to the complexity of the aesthetic and emotional perception process, obtaining such design information from the interacting context is challenging. This paper proposes an affective design approach based on the Kansei engineering (KE) method and a deep convolutional generative adversarial network (DCGAN) following the research trend of merging KE with computer science techniques in recent years. A case study of the social robot design is conducted to verify the effectiveness of this approach. Appearance aesthetic and emotional preference evaluations are adopted by the KE method first to identify the crucial features in two categories: (1) The physical features of the outer shape, head and color for aesthetics; (2) The emotional features of intelligent, interesting and pleasant for preference perceptions. Based on a manually created social robot image dataset, the DCGAN model is trained to automatically generate novel design images. Then several professional designers are involved to fine-tune the generated images in detail. The experimental results show that the newly designed social robots tend to obtain positive aesthetic and preference evaluations. Practically, such an affective design approach can help industrial design companies identify customers’ psychological requirements and support designers in creating new products innovatively and efficiently.     

A screening-based gradient-enhanced Gaussian process regression model for multi-fidelity data fusion

The prediction accuracy of multi-fidelity models can be enhanced by incorporating gradient formation. However, the computational complexity would increase dramatically as the number of design variables increase. In this work, a gradient-enhanced multi-fidelity Gaussian process model using a portion of gradients (PGEMFGP) is proposed. To be specific, a Bayesian Gaussian process regression model for multi-fidelity (MF) data fusion is developed, which incorporates high-fidelity (HF) and low-fidelity (LF) responses, as well as the corresponding gradients. A screening technique based on distance correlation is applied to select a portion of gradients of the low-fidelity model so that the modeling complexity can be greatly reduced. The merit of the proposed method is tested with six numerical examples ranging from 10-D to 30-D, as well as an aerodynamic airfoil case with 18 design variables. The proposed method is compared to two other existing gradient-enhanced Gaussian process-based models. It is shown that the modeling efficiency of the proposed model is dramatically improved compared to the original gradient-enhanced multi-fidelity Gaussian process model, while the loss of the prediction accuracy can be almost negligible. In consequence, it can be a promising approach for gradient-enhanced models dealing with multi-fidelity data.     

Research on construction and application of gene network model for form design based on consumer’s preference

With developments in science and technology, product innovation is the key to enterprise performance in the competitive market. In consumer demand-oriented modern product design, effectively combining product design with the esthetic perceptions and preferences of consumers is a problem that urgently needs to be addressed. However, the traditional design method is completely dominated by designers without user participation. The purpose of this study was to give designers an impetus for restructuring and upgrading a design. A gene network design model was developed with the advantages of a gene network and neural network. This allowed the target elements of a product design to be obtained using a nonlinear network. Finally, a case study was used to show the detailed procedure of the design model. To reveal the advantages of the proposed model, it was compared with other methods such as a gene-based design method, an emotional design method, and a fuzzy Kano design method. The results showed that the proposed model was more efficient and scientific, and provided consumers with a multidimensional evaluation system to determine the optimal design schemes.     

Design of structural optimal robust controller

The traditional robust controller is designed to meet the requirement considering both the disturbance and the plant uncertainty while the controller uncertainty is always neglected.The structural optimal robustness of the closed-loop system is proposed based on the analysis of the robust radii of both the plant and the controller.The subspace angle is introduced to measure the "distance" of two subspaces,and its metric is equivalent to the gap metric.The optimal robust controller based on gap metric is designed to control the rate of the line of sight for an electromechancial target tracking system.It is shown from simulations that the optimal robust controller with the biggest robust radius is superior on the ability of disturbance rejection,and high tracking performance when additive uncertainty exists compared with the robust controller with smaller robust radius.     

在线逆向拍卖供应商投标决策支持模型

为了解决基于分包算法的在线逆向组合多属性拍卖中供应商决策支持问题,通过对供应商投标行为的分析,以供应商收益最大化为目标,提出了基于计量经济学分包算法的分包确定模型以及供应商投标策略模型.通过算例演示表明,该方法可以有效地提高供应商的收益、降低拍卖过程中信息不对称所带来的负面影响.