Toyota Research Institute (TRI) today introduced a generative artificial intelligence technique to empower vehicle designers. Today, designers can take advantage of publicly available text-to-image generative AI tools as a first step in their creative process. With TRI's new technique, designers can incorporate initial design sketches and engineering constraints into this process, reducing the iterations required to reconcile design and engineering considerations.
"Generative AI tools are often used as inspiration for designers, but they cannot handle the complex engineering and safety considerations required in actual car design," said Avinash Balachandran, Director of the Interactive Driving Division. Humana (HID) from TRI, whose team worked on the technology. "This technique combines Toyota's traditional engineering strengths with the cutting-edge capabilities of modern generative artificial intelligence."
TRI researchers have published two articles describing how the technique incorporates precise engineering constraints into the design process. Constraints such as air resistance (affecting fuel efficiency) and chassis dimensions, such as ride height and cab dimensions (affecting handling, ergonomics, and safety), can now be implicitly incorporated into the generative artificial intelligence process.
The team combined principles of optimization theory, widely used in computer-aided engineering, with text-to-image-based generative artificial intelligence.
The resulting algorithm allows the designer to optimize engineering constraints while maintaining text-based stylistic cues for the generative AI process. New generative AI technique optimizes drag in successive iterations based on designer input.
As an example, a designer can request, via text prompt, a series of designs based on an initial prototype sketch with specific stylistic properties such as "sleek", "SUV-like" and "modern", while also optimizing a quantitative performance metric. In the research paper, the team focused specifically on drag. The approach can also optimize any other performance metrics or constraints inferred from a layout image.
"TRI is harnessing the creative power of artificial intelligence to empower automotive designers and engineers," said Charlene Wu, senior director of TRI's Human-Centered Artificial Intelligence (HCAI) Division, whose team collaborated with the Human Interactive Driving team in this project.