Toyota Research Institute (TRI) has announced a groundbreaking generative artificial intelligence (AI) technique set to empower vehicle designers. While publicly accessible text-to-image generative AI tools have become a source of inspiration in preliminary design phases, TRI’s latest innovation allows designers to integrate initial design sketches and crucial engineering constraints directly into this process. This advancement significantly reduces the iterative cycles typically required to harmonize design aesthetics with engineering practicalities when using Car Sketching Tools.
“Although generative AI tools often spark creativity for designers, they traditionally fall short when addressing the intricate engineering and safety considerations inherent in real-world car design,” stated Avinash Balachandran, director of TRI’s Human Interactive Driving (HID) Division, the team spearheading this technological leap. “Our new technique masterfully blends Toyota’s long-standing engineering prowess with the cutting-edge capabilities of contemporary generative AI, enhancing the functionality of car sketching tools.”
TRI researchers have published two detailed papers outlining how this technique seamlessly embeds precise engineering constraints within the design workflow. Parameters such as aerodynamic drag—critical for fuel efficiency—and fundamental chassis dimensions like ride height and cabin size—influencing handling, ergonomics, and safety—can now be inherently incorporated into the generative AI process used in car sketching tools. By connecting principles of optimization theory, widely utilized in computer-aided engineering, with text-to-image generative AI, the team has developed an algorithm that empowers designers to optimize engineering constraints while leveraging text-based stylistic prompts within their car sketching tools.
For instance, using these advanced car sketching tools, a designer can input a text prompt requesting a range of designs based on an initial prototype sketch, specifying stylistic attributes such as “sleek,” “SUV-like,” and “modern,” while simultaneously optimizing a quantifiable performance metric. In their research, the team specifically concentrated on aerodynamic drag. However, the methodology is adaptable to optimize various performance metrics or constraints derivable from a design image created with car sketching tools.
“TRI is effectively channeling the imaginative potential of AI to augment the skills of automobile designers and engineers,” commented Charlene Wu, senior director of TRI’s Human-Centered AI (HCAI) Division, whose team collaborated with the Human Interactive Driving team on this innovative project focused on car sketching tools.
By directly integrating engineering constraints into the design process through these AI-driven car sketching tools, Toyota aims to expedite and enhance the efficiency of designing electrified vehicles.
“Minimizing drag is crucial for boosting the aerodynamic performance of Battery Electric Vehicles (BEVs) to maximize their driving range,” emphasized Takero Kato, BEV factory president, Toyota Motor Corporation, highlighting the importance of advanced car sketching tools in modern automotive design.
Further technical insights into this pioneering technique are available in two papers:
Interpreting and Improving Diffusion Models Using the Euclidean Distance Function, F. Permenter, C. Yuan, 2023.
Drag-guided diffusion models for vehicle image generation, N. Arechiga, F. Permenter, B. Song, C. Yuan, 2023.
