GenAI Usage in Manufacturing and Its Benefits
Apart from producing multimedia, Generative AI has also been adopted to solve manufacturing-related issues. They include blueprint designing, production planning, supply chain management, product quality control, financial calculations, and even creation of new ideas and concepts.
While AI-powered manufacturing is a young industry, there are some advances in technologies developed with its help: a medicine for antibiotic-resistant bacteria, 5-kilonewton rocket engine TKL-5, Floor Planner for creating domestic architecture, and so on.
Generative Design and Its Use in Complex Geometries
It is debated that GenAI is crucial to the future of engineering, as it greatly complements additive engineering — when manufactured items are created with the 3D-printing process.
GenAI has a number of advantages in this regard: it allows working with complex geometries, enables customization to meet the end user’s expectations, as well as provides simulation capabilities crucial for testing and optimizing the product.
Some Tools and Applications for Generative Design
There are some proposed tools for GenAI engineering.
- GearFormer
GearFormer is a generative solution based on the combination of the:
- Transformer model.
- Estimation of Distribution Algorithm (EDA) for iterative sampling and output improvement.
- Monte Carlo tree search (MCTS) for balancing of multiple paths leading to the solution.
As a result, GearFormer is capable of synthesizing gear train units that also meet quality requirements.
- Intelligent CAD
CAD stands for Computer-aided Design, which is a tool for modelling, analyzing, modifying, and editing designs. AI-powered CAD can help solve a serious issue of informational incompleteness when it is basically impossible to predict how a device or contraption will behave after the detailed design stage is finished.
On the other hand, AI-assisted CAD can be capable of producing intricate simulations close to real life. Eventually, it will be much easier to modify prototypes, make suggestions on improvements, and so on.
Cases of Generative AI Solutions
Among the real-life solutions designed by AI it’s worthy to note:
- LEAP 71’s First Rocket Engine
A UAE-based company LEAP 71 developed a GenAI-engineering model dubbed Noyron. It’s based on a publicly available geometry kernel PicoGK and it can calculate various parameters, like thermal behavior, and also provide process parameters. In 2024, over the course of three weeks, Noyron developed an aerospike rocket engine that has a unique architecture and cryogene-based cooling system.
- Bionic partition 2.0
An airplane component named Bionic Partition 2.0 was designed by Airbus together with Autodesk. This component separates the passenger seat from the galley of the plane and is 45% lighter than partitions with regular design, while also equally sturdy.
It is estimated that Partition 2.0 — produced with 3D printing — can considerably lower manufacturing costs and reduce CO2 emissions by half a million tons.
- Creating Metamaterials
Metamaterials display “smart” characteristics due to their unique structure and anatomy consisting of various ordinary materials. A novel solution was designed to modulate metamaterials: it is based on a parameter-free design strategy that employs a latent space with clusters of unit cell topologies that share related properties.
- Designing and Creating Soft Robots
Soft robotics is a subset of traditional robot science that focuses on roboticized machines consisting of soft, flexible structures. It is suggested that a generative model based on latent diffusion (SDFusion) can effectively learn data distribution from studying robot actuator designs.
- Protein Engineering
Another GenAI-oriented concept is capable of amino acid sequence construction that in turn can form proteins. A combination of Autoregressive models, Variational Autoencoders, Generative Adversarial Network, and a diffusion model dubbed EvoDiff show promising potential in the field.
Antispoofing
