Conference Agenda

The widespread use of large multimodal models (LMMs) and the ever-growing interest in their application across all fields raise questions about their reliability in areas such as mechanical design and CAD-related workflows. Before entrusting them with complex operations, it is important to establish whether the initial stage of interpreting drawings is robust enough and to evaluate their ability to correctly recognise and classify all the elements inside drawings. This paper describes an evaluation protocol implemented for three common LMMs in analysing mechanical drawings taken from textbooks and realised by means of CAD, as well as drawings created by hand by undergraduate students. The intention is not to determine which one is best, but rather to test their robustness in the initial stages of reading technical drawings. Each model is prompted to extract elements according to five categories: Views, Sections, Features, Dimensions, and Annotations. The results obtained are then compared with a ground truth, defined by experts, across four dimensions: Presence, Text, Type, and Context. This allows the overall results to be evaluated, as well as separate analyses of the five categories to consider factors such as false positives, omissions, element splitting, and uncertainty handling. The outcomes suggest that LMM-based drawing analysis should be reviewed using multidimensional protocols, as aggregate scores could mask differences in how models recognise and structure technical information.

Axiomatic Design detects technical contradictions through

the analysis of the Design Matrix that makes evident the dependencies

between functional requirements (FR) and design parameters (DP). For

this purpose, it is required that the set of FR is complete; a missing re-

quirement leaves no cell in the matrix, so the conflict to be revealed stays

hidden. Requirements cannot be expressed because they are taken for

granted, overlooked, or simply unknown, making specification complete-

ness hard to guarantee. This paper proposes a method in which a Large

Language Model (LLM) assists the designer during each decomposition

step by suggesting sub-functions, requirements, and parameters, includ-

ing elements that the designer may not have considered. In the proposed

approach, the designer validates each suggestion to ensure the final set

remains under the designer’s control. The method is demonstrated in an

emblematic case study, the western hand saw, where evident contradic-

tions can be revealed. This case is compared with the oriental hand saw,

which resolves some of these contradictions. A verification across three

open-weight LLMs shows that the models reliably replicate the expert’s

decomposition and requirement discrimination, correctly distinguishing

the two concepts, while the independent filling of the design matrix is

the most fragile step; the designer’s judgment remains essential.

Textile metamaterials are fabrics whose mechanical or informational properties derive from how their structure is arranged in space rather than from the fibre alone. Knitting is a scalable option: an additive manufacturing technique that programs properties locally and already has industrial adoption. What blocks this passage is the design workflow: conventional knitwear CAD treats the step from design intent to manufactured object as a technical translation, limiting designers’ control over performance, manufacturability, and iteration. The paper reports a practice-based case study from the knitdesign.polimi research group at Politecnico di Milano, asking how a custom digital environment can reconnect image, structure, machine logic, and garment construction within an integrated product-development process. Developed through AI-assisted coding and tested on an adversarial knitted textile, the redesign frames software as part of the material and product design space. Prototyping a knit-to-shape garment shows how adaptive, designer-led CAD can support industrial scaling of textile metamaterials and reconfigure collaboration between design, engineering, and technical expertise.

Composite materials, using renewable natural fibers and biomass resources as reinforcing phases, offer a sustainable alternative to petroleum-based composites by reducing dependence on fossil resources and enabling bio-degradability. However, the scale-up of these materials from laboratory in-novation to full-scale industrial implementation is hindered by significant challenges, including the variability of waste feedstocks, limited matrix-filler interfacial stress-transfer, and processing limitations. This paper pro-poses a comprehensive, eight-action composite design protocols specifical-ly tailored for upcycling pre-consumer textile and agro-industrial waste. By systematically integrating waste characterization, matrix selection, possible surface treatment, and processing parameters optimization, the protocol ad-dresses the heterogeneity inherent in industrial byproducts such as cotton offcuts, denim, rice husks, and hemp shives. The study includes/gathers findings from academic literature and industrial case studies to map the life cycle of these materials, from resource acquisition to end-of-life options. The proposed framework moves beyond a linear "drop-in" approach, offer-ing a flexible, iterative methodology that aligns material properties with application requirements. This work contributes to the broader discourse on circular economy strategies, demonstrating a viable pathway for trans-forming diverse industrial waste streams into high-performance, scalable composite materials for automotive, construction, and packaging applica-tions.

The integration of Extended Reality (XR) technologies, Virtual Reality (VR), and Augmented Reality (AR), into the early stages of new product de-velopment is reshaping how designers and engineers generate, communi-cate, and evaluate product concepts. Coupled with rapid prototyping (RP), XR enables low-cost, iterative, and immersive testing of product ideas be-fore committing resources to physical production, supporting more in-formed decisions on form, function, and user experience (UX). This paper investigates how XR and RP can be combined into a flexible, increasingly accessible evaluation pipeline for new product ideas. After reviewing the state of the art on XR-based evaluation methods and on the integration of rapid prototyping within immersive design processes, the paper proposes a theoretical framework that positions XR and RP not as isolated tools but as complementary, modular instruments that can be configured according to the fidelity, interactivity, and UX dimensions required at each development stage. The framework is discussed through two case studies in which im-mersive prototypes and rapidly fabricated physical mock-ups were devel-oped for evaluating alternative product concepts with end users. Prelimi-nary findings indicate that the combined use of XR and RP improves the richness of user feedback and shortens iteration cycles, suggesting their growing role as standard instruments in design practice.

Understanding, internalising, and applying the Design Engineering (D&E) process is rarely straightforward. In D&E education, process representations act as shared artefacts that support communication between students and faculty. However, they often struggle to capture the complexity, dynamism, and situated nature of D&E practice. Previous work addressed this challenge through the DElta model, which provides a richer representation of the D&E process within the Master's in D&E at [Institution]. Building on this work, this paper investigates how the D&E process is understood, expe-rienced, and represented across different educational roles. Using a collec-tive autoethnographic approach, four authors—a student, a teaching assis-tant, and two faculty members—independently reflected on their experienc-es before engaging in a cross-case analysis. The analysis revealed three par-allel interpretative trajectories that describe how educational experience progressively reshapes the meaning attributed to the process, the relation-ship established with it, and its representation. Far from proposing a new process model, the study offers an educational interpretation of existing process representations, showing how they become reflective artefacts through which learners and educators interpret their own development with-in D&E.

Generative AI (GenAI) is increasingly entering engineering and design education, yet its integration into structured creative processes remains insufficiently understood. This paper presents an instrumented, sprint-based workshop that teaches engineering students to integrate creativity and GenAI into design practice while making their tool use observable across the process.

The workshop was implemented with three cohorts: first- and second-year Video Game Engineering students and first-year Master's students in Mechanical Engineering (Marche Polytechnic University). After a brief introduction to GenAI and creative processes, participants addressed comparable design briefs through seven sprints organised into four phases: Preparation, Insight, Evaluation and Elaboration. Students freely chose their preferred GenAI tools; processes were documented via Miro boards, AI conversation logs, final artefacts, and post-workshop questionnaires.

The paper describes the workshop framework and presents a preliminary qualitative analysis of 30 documented design processes, using Miro boards for case summaries and questionnaires and conversation logs as contextual evidence. Cross-case comparison showed GenAI use was strongly phase-dependent: almost universal during Elaboration, mainly as a visualisation tool, but uncommon during Evaluation. Process Governance was Human-led in 16 cases, Mixed in 8, and AI-driven in 6. The workshop offers a transdisciplinary framework for teaching and observing human--AI co-creative design processes.