AI in industrial design: incoming revolution?

AI in industrial design: incoming revolution?

Keywords: generative design, AI, design, industry

A chair designed by artificial intelligence? Yes, it is now possible, as recently demonstrated by Philippe Starck and Kartell, with their new chair simply called “A.I.”.

Artificial intelligence is a source of innovation in many various fields nowadays, and it may soon create a small revolution in the world of industrial design. If you work in the industry, if you’re a designer or if you’re an AI enthusiast, you might be interested in learning more about this innovation called “generative design”.

What is a generative design?

Chair “A.I.” (2019) by Philippe Starck and Kartell, in collaboration with Autodesk.

The technology used by Philippe Starck to design this chair is called generative design. Generative design is an iterative process, driven by an algorithm, which aims at creating the best possible shape, given initial requirements.

So far, it doesn’t sound much like artificial intelligence. Indeed, industrial companies did not wait for the development of new artificial intelligence methods to start using honeycomb structures, for example, or to create complex shapes in order to reduce the weight of a given structure.

There are already a lot of existing optimization algorithms in the industry; this is called topology optimization. Topology optimization consists in optimizing the quantity of material for a shape with a given mechanical stress configuration. Such algorithms remove material from the shape, depending on mechanical simulation results.

However, generative design is way more sophisticated than topology optimization: it is literally able to generate new shapes (as you may have already guessed). The algorithm won’t just reduce the quantity of material of the initial shape; it will instead explore several design options and select the best ones.

Especially, the algorithm will produce shapes that can be dramatically different from the initial one. Forms that maybe the designer did not think of at all.

Generative design algorithm results: as illustrated in this picture, you can see that the algorithm tries diverse design variations to match user requirements (General Motors and Autodesk).

Generative design could dramatically speed up your industrial design process and result in very sophisticated shapes.

Generative design is not yet a very mature technology and is probably used today mostly for relatively simple objects (like “A.I.” chair) or for experimenting with this technology. However, we can already foresee three main advantages of this technology:

Generative design is very flexible.

Contrary to topology optimization methods, generative design algorithms can take into account many more parameters, like, for example, the desired manufacturing method. This is really critical, as usually, the result given by topology optimization methods should be re-design afterward to suit the desired manufacturing method.

Generative design tends to better process integration.

As more parameters are taken into account, the result given by a generative design algorithm is much more likely to be near to ready for production. Ideally, all steps can be done by the same person, 3D files won’t have to come and go between several departments anymore, thus speeding up the whole process.

Generative design can achieve very high optimization.

1. Indeed, generative design can explore design options not even considered by the designer. Thus, it can potentially achieve very high optimization or produce very complex structures in a short period of time.

Generative design will enable specialists to focus on their core skills and will reveal the full potential of additive manufacturing methods.

Despite a very high level of autonomy, it is still necessary to define part requirements, like the functional surfaces, the mechanical stress limits, or the material used.

Just like finite element methods, the generative design represents a new tool for designers, in order to enable them to create more complex and optimized shapes while spending less time on repetitive tasks and calculation.

However, given the potentially high complexity of shapes, the use of conventional manufacturing methods will probably create a new bottleneck for the whole development process, thus limiting the potential of generative design.

Nonetheless, additive manufacturing methods have considerably improved in the past decades. The combination of both additive methods and generative design might create a powerful synergy.

Indeed, contrary to conventional methods, the time of manufacturing with additive methods doesn’t depend so much on the shape complexity, but instead on the quantity of material. So, it’s a perfect match with shape complexity and material optimization allowed by generative design.

As an illustration, Airbus recently developed a new plane partition (a partition is a separation wall inside of plane cabins) using generative design. The results are impressive:

• This new partition is 45% lighter (it represents a fuel saving of about 3180kg per year per partition!)
• When using an additive manufacturing method, the partition needs 95% less of raw material.

• plane partition designed by Airbus (more details here)

This use case of artificial intelligence applied to industrial design is a perfect example illustrating how artificial intelligence can bring innovations in various fields. Far from the science fiction scenario, we are talking here of new powerful tools, already demonstrating concrete results, which can help us create more complex and sophisticated products in a short period of time. Just like Philippe Starck said concerning his chair design process: “From elsewhere, a new world opens up to us. Unlimited.”