The Toyota Research Institute (TRI) has introduced a cutting-edge generative AI tool that will aid in the imagination of automobile designers. The software provides textual instructions to designers, who may then produce design concepts with certain style features like \”sleek,\” \”SUV-like,\” and \”modern.\” Designers may also use quantitative performance indicators to optimise the first draught of a prototype.
The business claims this new technology will provide designers more freedom to innovate while still guaranteeing high-quality, time-saving design process.
The proposed approach may be integrated into various text-to-image based generative AI models, as described in two publications released by TRI researchers. The method behind the tool\’s picture creation is explained in these publications.
The group used optimisation theory, widely employed in computer-aided engineering, with generative AI based on text to produce a new visual. As a consequence, the method enables designers to maximise their freedom under technical limits without sacrificing the integrity of their generative AI\’s textual style cues.
Improving the Creation of Images
Vehicle restrictions like drag, which has a direct influence on fuel economy, and chassis characteristics like ride height and interior dimensions, which have an impact on handling, ergonomics, and safety, may now be implicitly integrated by designers to improve image creation.
\”Current text-to-image generative AI tools primarily focus on adhering to the designer\’s text-based stylistic guidelines when generating potential images,\” Avinash Balachandran, head of Toyota Research Institute\’s Human Interactive Driving (HID) branch, told VentureBeat. As we put it, \”our technique allows users to explicitly incorporate and optimise over-engineering constraints like drag or ride height while generating images that adhere to the designer\’s stylistic guidelines.\”
Balachandran said that such methods, which strike a better balance between aesthetic and technical considerations, might hasten the development of novel designs.
Adding that \”any designer can use generative AI tools for inspiration, but these tools cannot handle the complex engineering and safety considerations that go into actual car design,\” he said that this was because of the limitations of the tools themselves. Our designs have to adhere to technical standards if we want to produce automobiles that are both safe and dependable. By imposing limits on generative AI, the designer may give the algorithm\’s outputs more structure.
Generative artificial intelligence for improved automobile layout
According to Balachandran\’s interview with VentureBeat, the project started out around a year and a half ago, inspired by the development of text-to-image generative AI technologies that enable users to submit a prompt and get a picture that conforms to the specified style direction.
He said, \”Our vehicle designers in Toyota told us how one of the challenging parts of the design process for them was to come up with inspiration for new designs.\” They also mentioned how challenging the iterative process between design and engineering was in order to create a design that is both visually appealing and functionally sound.
Balachandran claims that designers and engineers often come from a variety of educational and cultural backgrounds and approach problems in unique ways. As a consequence, the majority of the time a designer generates a design, it falls short of the original technical criteria and requires extensive cooperation with the engineering team to arrive at an ideal solution.
The inherent friction between design and engineering, in addition to the iterative nature of the design process, both contribute to its length.
The idea for this method and its accompanying tools came from a desire to \”shorten that iteration loop between engineering and design,\” as Balachandran put it.
Combining many sources of information
According to Toyota, the notion of a \”AI assistant\” that suggests new ideas by using many varied data sources was a popular one during brainstorming sessions with designers. This gave rise to the concept of building a design-creation tool that incorporates generative AI and several streams of data, such as technical limitations.
Charlene Wu, senior director of Toyota Research Institute\’s Human-Centered AI (HCAI) division, told VentureBeat that the company\’s designers were able to better focus on identifying constraints and important stylistic aspects of the design thanks to the incorporation of generative AI technology. We think our solution will free them up to spend more time on the parts of the creative process where they can really shine.
Where does Toyota go from here?
While the technology is still in the testing phase, the business has said that it is working with Toyota departments to include the tool into their car design and development processes. TRI has pledged to keep doing studies that improve people\’s and society\’s standard of living.
\”The hope is that by using this tool, vehicle designers around the world can expand the power of design ideas while at the same time drastically improving the speed of design development,\” said Balachandran. \”Generative AI is a potent new tool, and we\’re exploring how to leverage it responsibly across our many research areas so that it can amplify people.\”