Enabling Generative Design Tools with LLM Agents for Mechanical Computation Devices: A Case Study

TL;DR

The Generative Design Tool (GDT) better understands the capabilities and limitations of new technologies, proposes diverse and practical applications, and suggests designs that are technically and contextually appropriate.

Abstract

In the field of Human-Computer Interaction (HCI), interactive devices with embedded mechanical computation are gaining attention. The rise of these cutting-edge devices has created a need for specialized design tools that democratize the prototyping process. While current tools streamline prototyping through parametric design and simulation, they often come with a steep learning curve and may not fully support creative ideation. In this study, we use fluidic computation interfaces as a case study to explore how design tools for such devices can be augmented by Large Language Model agents (LLMs). Integrated with LLMs, the Generative Design Tool (GDT) better understands the capabilities and limitations of new technologies, proposes diverse and practical applications, and suggests designs that are technically and contextually appropriate. Additionally, it generates design parameters for visualizing results and producing fabrication-ready support files. This paper details the GDT's framework, implementation, and performance while addressing its potential and challenges.

Figures (12)

• Figure 1: Building an interactive sitting posture correction chair using the Fluidic Computation Kit involves: a) Selecting the basic components; b) Assembling operators by wiring the components; c) Constructing the circuit with operators based on the logic; and d) Preparing and integrating input/output airbags with the circuit into the chair. (Permission granted from the authors)
• Figure 2: The GDT’s overview. Sections a-g constitute the backend, which includes LLM agents and conventional algorithms. Sections h-m represent the frontend, utilized by users interacting with the GDT. The Consultant agent (a) and the first tab of the GUI (h,j) are dedicated to assisting with the design ideation and detailing phase. The remaining components are focused on the solution implementation phase. Some text within the GUI has been enlarged to enhance readability. The heat-sealing pattern for the output module (m) will be provided when the type of output is shape-changing, and the shape is associated with an inverse design algorithm.
• Figure 3: The user flow for designing with Fluid Computation GDT includes: (1) Greeting the Consultant and freely asking for explanations to learn about FCI. (2) Setting the design goal with the help of ideation. (3) Defining the input, output, and computation modules with the Consultant's guidance and recommendations. (4) Confirming the design definition and previewing the generated fluid computation system; clicking the input module to see the animated demonstration of control logic.
• Figure 4: Categorization of the one hundred design goals proposed by the GDT, based on their application scenarios.
• Figure 5: Rating result of the one hundred design goals.