Application of Feature-Learning Methods Toward Product Usage Context Identification and Comfort Prediction

Abstract

Usage context is considered a critical driving factor for customers' product choices. In addition, physical use of a product (i.e., user-product interaction) dictates a number of customer perceptions (e.g., level of comfort). In the emerging internet of things (IoT), this work hypothesizes that it is possible to understand product usage and level of comfort while it is “in-use” by capturing the user-product interaction data. Mining this data to understand both the usage context and the comfort of the user adds new capabilities to product design. There has been tremendous progress in the field of data analytics, but the application in product design is still nascent. In this work, application of feature-learning methods for the identification of product usage context and level of comfort is demonstrated, where usage context is limited to the activity of the user. A novel generic architecture using foundations in convolutional neural network (CNN) is developed and applied to a walking activity classification using smartphone accelerometer data. Results are compared with feature-based machine learning algorithms (neural network and support vector machines (SVM)) and demonstrate the benefits of using the feature-learning methods over the feature-based machine-learning algorithms. To demonstrate the generic nature of the architecture, an application toward comfort level prediction is presented using force sensor data from a sensor-integrated shoe.

Assessing Impact of Sensors and Feature Selection in Smart-Insole-Based Human Activity Recognition

“…The challenge with this approach is determining the appropriate window size, because a window size that is too small will not accurately represent all of the activities involved, and a window size that is too large will include multiple activities in one window, impacting the final classification. Typically, empirical choices are made in the literature to set the window size based on the researcher's experience or utilising existing state-of-the-art studies as a reference, for example, Merry et al [16] proposed a window size of 0.5 s, Truong et al [23] a window size of 10 s, and Ghosh et al [24] a window size of 60 s.…”

Application of autoencoders in cyber-empathic design

“…A limitation of the CED framework is that information representing user–product interaction is extracted manually from raw sensor data (i.e., features from raw sensor data as in Figure 1). This reliance on pre-designed sensor features is limited by the extent of the domain knowledge of the designers and typically requires significant knowledge to identify relevant features for different contexts (Ghosh, Olewnik & Lewis 2017 b ). The investigation undertaken in this work is focused on an approach to automate extraction of measurement features from raw sensor data to be mapped on the model of latent constructs representative of user perceptions (Figure 1).…”

Design of a Deep Post Gripping Perception Framework for Industrial Robots

“…This can be realized by the use of computers in tasks that require knowledge, perception, reasoning, learning, understanding and similar cognitive abilities, also called artificial intelligence [9]. Machine learning, which is the use of learning agents that improve their performance on future tasks after making observations about the world [10], and mainly deep learning, meaning to learn using artificial neural networks with hidden layers, has known a wide application spectrum [11] and an important success recently [12]. The use of this artificial intelligence ability increases the autonomy of robots and their ability to adapt to new conditions, which allow their usage in novel applications [13][14][15][16].…”