Page 14 - IJAMD-1-3

International Journal of AI for
Materials and Design
Smart cockpit design with generative models

Figure 1. CockpitGemini: The proposed personalized design framework integrating generative model-based multi-agent systems and human digital twin
technologies in an immersive environment

agent monitors and analyzes user behavior data from which can be inferred from partial or comprehensive
HDT to further discover user preferences, providing a physiological data. For instance, Tanwar et al. leveraged
48
foundation for design optimization and iteration. The hybrid deep learning models to process multi-modal
user receives the design results from the human-in-the- physiological data (heart and electrodermal activity) to
loop multi-agent collaboration process, visualized in an predict users’ stress states. In addition, biological data refers
immersive environment that enhances the sense of realism to the data types at the organ or cell level, primarily applied
and improves the overall UX. 45 in smart healthcare and medicine. Given the complexity
HDTs can significantly enhance real-time status and variety of data for building HDT, data management
monitoring, design verification, performance evaluation, plays a crucial role in transmitting, storing, integrating,
and iterative design within virtual environments, and managing heterogeneous data within databases.
representing a critical component of immersive design. This is often achieved through cloud computing, wireless
The development of HDT encompasses four primary sensor networks, and database technologies. Subsequently,
stages: data acquisition, data management, data analytics, data pre-processing is performed to eliminate redundant
and human modeling. Initially, data acquisition forms the and irrelevant data, involving processes such as data
cornerstone of HDT, encompassing four key dimensions: filtering, augmentation, standardization, and feature
49
physical data, physiological data, psychological data, 10,34 dimensional reduction. Afterward, data analytics
and biological data. Physical data refers to the user’s involves multi-dimensional user individual data using
46
body features, such as anthropocentric data (e.g., standing advanced algorithms to discover the internal mechanism
height, seating height, etc.) obtained through RGB-D of the human body and establish an individual HDT that
47
cameras, physical movements captured by optical is highly compatible with personal characteristics. Finally,
motion cameras, and scanning systems. Physiological data human modeling entails the digital representation of
involves the digital representation of various biological physical counterparts, intuitively reflecting their behaviors
signals measurable from the human body, including eye and characteristics. The variety of HDT models aligns
movement data, electrocardiogram, electromyogram, body with the obtained data dimensions, including physical,
temperature, and acceleration. These data provide insights psychological, and biological models. The physical model,
into an individual’s physiological state and responses to driven by physical and physiological data, can reflect
different stimuli or activities. Psychological data reflects human body appearance, motion, ergonomics, and
users’ mental status, such as emotion, fatigue, and stress, performing basic activities (e.g., walking, jumping) in a

Volume 1 Issue 3 (2024) 8 doi: 10.36922/ijamd.4220