By Design: How AI is Reinventing Product Creation

May 9, 2017 by Guillaume Polaillon

Deep learning is coming to 3D design. Just ask Dassault Systèmes.

The company’s 3DExperience Platform is used for everything from Tesla cars and Boeing airplanes to Procter & Gamble consumer products. Now, Dassault Systèmes is amping up 3D design and simulation with deep learning and GPUs.

“What people have done with deep learning for image and speech recognition, we’re doing to transform how products are designed and experienced,” said Patrick Johnson, vice president for corporate research at the 3D modeling software company.

How AI Helps Designers

Traditionally, designers and engineers had to start each new product from scratch.

With NVIDIA GPUs and deep learning, Dassault Systèmes will soon be able to harness the history of previous designs. The system’s 3Dbots use this knowledge to predict and propose a whole new range of designs in a exploratory innovation space. Romain Perron, R&D web apps and services director for the company’s flagship CATIA brand, explained these cognitive augmented design 3Dbots in more detail at  the GPU Technology Conference,

“From there, you just select the ones you want for faster and better design, leveraging know-how and past designs,” Johnson said. Engineers can use the company’s software to plan the ideal manufacturing process and see how their design will function in the real world.

Shipmaker Meyerwerft used Dassault Systèmes' GPU-powered 3D design software to create this ship, which has 300,000 different parts.
Shipmaker Meyer Werft used Dassault Systemes’ GPU-powered 3D design software to create this ship, which has 300,000 different parts.

3D Design at GTC

The company was out in force at GTC with sessions covering topics such as  how it’s using deep learning, virtual and augmented reality, and NVIDIA Iray rendering software to advance 3D modeling:

To learn more about how AI computing is transforming industries, watch for replays of these and other talks from the GPU Technology Conference, May 8-11, in Silicon Valley.