Generative Assembly Line Design: Optimizing Task assignment, Equipment Selection and Balancing

This paper introduces Generative Assembly Line Design (GALD), a novel framework leveraging artificial generative intelligence to optimize task assignment, equipment selection, and line balancing in assembly line design. GALD integrates Variational Autoencoders (VAEs) and Generative Adversarial Networks (GANs) to address the complex interdependencies of assembly line design challenges. The framework minimizes total lifetime costs, encompassing worker wages, material handling, and equipment expenses, while adhering to stringent ergonomic and safety constraints. The methodology emphasizes a unified optimization approach rather than isolated problem-solving modules, with explicit consideration of ergonomic factors and demand-driven cycle times. The proposed GALD framework operates in two synergistic stages. The first stage utilizes VAEs to explore a diverse solution space, generating latent representations of feasible configurations guided by life cycle cost profiles. The second stage employs GANs for iterative refinement, where a generator proposes improved assembly line configurations, and a discriminator evaluates them for cost-effectiveness and compliance with operational constraints. This two-stage generative process ensures cohesive and efficient task-equipment assignments and balanced workloads. By synthesizing generative algorithms and industrial engineering principles, GALD offers a robust and cost-efficient solution to modern assembly-line design This study provides a potential for future extensions to incorporate human-robot collaboration, aesthetics, and sustainability metrics into the line design.