Toyota Motor Research Institute launches generative AI car design tool

The newly launched innovative generative artificial intelligence tool will enable designers to explore their creativity while ensuring efficient and effective design development, Toyota Motor Research Institute said.

The Toyota Motor Research Institute researchers also published two papers describing how the developed techniques can be integrated into other text-to-image-based generative AI models. These papers illustrate the tool's image generation process.

The team combines optimization theory widely used in computer-aided engineering with generative artificial intelligence based on text-to-image. The algorithm allows designers to optimize the content of engineering constraints while retaining a text-based style guide for the generative AI process.

Enhanced Image Generation

Designers at Toyota Motor Research Institute can now incorporate vehicle constraints such as drag, which directly affect a vehicle’s fuel efficiency, as well as the chassis, which affects handling, ergonomics and safety. Dimensions such as ride height and cabin dimensions are integrated to enhance image generation.

Avinash Balachandran, head of the Human Interactive Driving (HID) Division at Toyota Research Institute, said: “Current text-to-image generative AI tools primarily focus on adhering to the designer’s text-based style guidelines when generating latent images. Our technology allows users to explicitly incorporate and optimize over-engineered constraints (such as drag or ride height) while producing images that adhere to designer style guidelines."

Balachandran said this technology can be faster and more precise Effectively balance trade-offs between aesthetics and engineering to speed the creation of new designs.

He added, “Any designer can use generative AI tools for inspiration, but these tools cannot handle the complex engineering and safety considerations of actual car design. To build safe and reliable vehicles, we The design must meet engineering requirements. Essentially, adding constraints to generative AI actually provides users with the ability to add guidance trajectories to their generative designs."

Optimizing vehicle design through generative AI

The project began about a year and a half ago, Balachandran said, thanks to advances in text-to-image generative artificial intelligence tools that allow users to input prompts and, in response, generate text that is consistent with the style guidance provided. Image.

Providing inspiration for new designs is one of the most challenging parts of the design process for our automotive designers, he explains. Achieving the required aesthetics, engineering performance, and safety measures is very challenging with an iterative process between design and engineering. ”

According to Balachandran, designers and engineers often have different backgrounds and mindsets. Therefore, when a designer creates a design, it often doesn’t meet the initial engineering requirements, resulting in a lot of work with the engineering team. Collaborate to arrive at the best solution.

This iterative process, coupled with the inherent tension between design and engineering, helps extend the duration of the design.

Balachandran said: " This technology and these tools are inspired not only to inspire creativity, but also to shorten the iteration cycle between engineering and design. ”

Integrating disparate data streams

Toyota said that during ideation sessions with designers, the concept of an “artificial intelligence assistant” that would work by leveraging multiple Different data streams suggest new designs. The idea is to integrate generative AI into tools that incorporate various data streams, including engineering constraints, to produce innovative designs.

Charlene Wu, senior director of Toyota Research Institute Artificial Intelligence (HCAI) Division, said: “By integrating generative AI technology, we found that designers were able to focus on identifying design constraints and important stylistic aspects while ensuring that engineering constraints were met. We are confident that our tools allow users to focus more on and get greater value from the design process they love most. ”

What’s next for Toyota?

The company announced that while the technology is currently in the research phase, they are working with teams within Toyota to integrate the tool into in the vehicle design and development process. TRI, the Toyota Research Institute, has expressed its commitment to ongoing research aimed at enhancing both personal and societal quality of life.

We hope that by using this Tools that allow automotive designers around the world to expand the power of design concepts while dramatically increasing the speed of design development. Many of our researchers are studying how to responsibly utilize generative AI, a powerful new tool. ”

The above is the detailed content of Toyota Motor Research Institute launches generative AI car design tool. For more information, please follow other related articles on the PHP Chinese website!