CAD More Room for Creativity with AI

Related Vendors

Dassault Systèmes Deutschland GmbH

Wuxi InfiMotion Technology Co., Ltd.

The "C" in CAD now once again stands for "Creativity." An overview of how machine learning and language models can assist designers in their daily tasks.

The rapid rise of Computer-Aided Design (CAD) in the mid-1980s was followed by a long-lasting decline starting in the late 1990s. With the emergence of initiatives around data management and PLM (Product Lifecycle Management), far less attention was paid to the computer-aided creation of geometries. License revenues declined, which was, of course, also due to increasingly affordable hardware.

However, the event "Catia Creative Design & Engineering Days" at the Science Congress Center on the campus of the Technical University of Munich (TUM) in Garching demonstrated that CAD has now returned to the forefront of computer-aided engineering sciences. One reason why CAD is no longer dismissed as a "commodity" is that it has essentially evolved into CA(AI)D—CAD enhanced by artificial intelligence.

CAD as Part of a Larger "Universe"

The conference by Dassault Systèmes discussed the necessities for transitioning from Catia V5 and the high-end surface modeling tool Icem Surf to 3D-Experience Catia. However, the event also reflected an overall generational shift from document-centric to platform-based approaches. According to leading system providers, CAD should be seen as part of a much larger concept of an "(Industrial) Metaverse" (Siemens, Nvidia, PTC) or "3D Univ+rses" (Dassault Systèmes). In this blend of virtual and real realities, users interact via digital avatars with digital twins of machines, other objects, and within an emulated environment.

But does this vision truly hold potential for industrial practice? Or is it merely a fantasy? "3D modeling alone is no longer sufficient today. If you want to manage the complexity in product development, you have to think and model in systems. This includes requirements, behavior, product architectures, systems of systems within a platform," emphasized Olivier Sappin, CEO of Catia.

It is these various perspectives that must be considered in a holistic modeling of digital twins. Take the digital representation of the so-called "Welcome Sequence" as an example, where a vehicle greets the driver: this involves not only the realistic 3D modeling of the (mechanical) instrument panel with all its materiality but also ray tracing of the appealingly illuminated cockpit (electronics and software). Additionally, other facets of the human-machine interface come into play, such as how the driver interacts with the car as a whole ("Driver Experience").

It Has to Be Faster

In addition to all the added possibilities of modeling realistic experiences with the virtual product, the so-called front-loading must not be overlooked. Even when simulating new scenarios, the time-to-market must be kept as short as possible. Among experts, regardless of the system provider chosen, discussions are candid and seriously address the possibilities of halving product development times. Chinese automakers, in particular, are leading the way.

In response, Dassault Systèmes relies, among other things, on the close integration of CAD and Modsim. The latter is a combination of simultaneous modeling and simulation, enabling designers to verify their designs themselves to avoid long processing times. Modsim, as an essential component of the "Democratization of Simulation"—the prevalent term for this—eliminates the need to wait for simulation engineers and their sharp expertise by providing access to pre-validated, automated CAE processes.

Another booster in terms of speed is the use of machine learning or artificial intelligence (AI). Dassault Systèmes has used the entire training material to train an AI. Starting with this year's release, every Catia application will be accompanied by the chatbot Aura. If questions arise, they will be answered contextually by accessing the training material. Another example of integrated AI is that, in a manufacturing drawing derived from the 3D model, the placement of dimensions is automated—with the correct arrangement of measurements. The AI was trained on old drawings for this purpose.

A New Job Profile Emerges

Clearly, the role of the designer is undergoing a profound transformation. Already today, a "geometry skeleton" can be initiated, leaving the detailing to the computer. In the first step, it is no longer the human but rather the bill of materials and integrated simulation techniques that dominate the process.

How dramatic the impact of "CAAID" can be is evident in the example of topology optimization: based on applied forces and boundary conditions, an initial topology is created—this is nothing new. Previously, however, many manual tasks would follow to arrive at a CAD model suitable for manufacturing. Nowadays, the computer can autonomously optimize the geometry according to manufacturing-relevant criteria and refine the surfaces AI-supported to a sufficient quality.

The acronym "Catia" stands for "Computer-aided three-dimensional interactive Application." The CAD/CAM/CAE application was first launched in 1977 by Dassault Aviation and further developed from 1981 by the IT spin-off of the aircraft manufacturer, Dassault Systèmes. Icem Surf (originally VW Surf) was developed in the 1980s for high-quality 3D surface modeling ("Class A Surfaces"). After an eventful company history, it has been part of Dassault Systèmes since 2007.

With version 5 of Catia, parametric modeling was introduced. The (highly successful) alternative to this in the 1990s was Pro/Engineer (now Creo) by PTC and later Unigraphics (now NX). All of these authoring systems succeeded in integrating the predecessor of AI applications, KBE (Knowledge-based Engineering, also known as knowledge management), into their mechanical CAD environments. KBE and the associated rule-based feature creation are key prerequisites for AI applications in CAD. Today, we stand on the threshold of AI-based CAD systems acting independently and delivering results that can be surprisingly efficiently utilized further.

E-CAD Benefits as Well

Incidentally, electrical engineering is also currently making a significant leap forward. At the recent Hannover Messe, Eplan presented new use cases for the implementation of Large Language Models (LLMs) based on Microsoft OpenAI Services. One use case revolves around the Eplan Copilot. For example, the prompt "What must I consider to comply with the IEC 81 346 standard?" provides this assistant with a targeted workflow for standards-compliant design.

The advancements in CAD are remarkable, and more and more routine tasks will disappear under the influence of AI. However, the designer cannot afford to rest on old habits.