AI-Powered Solutions for Toyota EV Engineers: Conquering Design Challenges
A new AI technique will help engineers and designers create vehicles that meet a variety of conflicting requirements.
The design of consumer products invariably requires a compromise of several factors and concerns. This is especially true in the automotive world, where interior space and comfort, safety, autodynamics and mechanical efficiency, manufacturability, and sustainability must be combined into a stylish design that retains a family resemblance to the company’s prior efforts.
Toyota Research Institute unveils generative AI for vehicle design. Image courtesy of XD: Experimental Design Studio of Toyota
The compromises made by automobile designers are even more complicated today than in the past. The entire industry is undergoing a wholesale change to electrification, and the design and development of electric vehicles (EVs) have changed many of the rules of the game. Today’s designers need new tools to meet the challenges, and at least one company is looking at putting artificial intelligence (AI) in its toolbox.
Generative AI Electric Vehicle Design
Toyota Research Institute (TRI) has unveiled a generative artificial intelligence (AI) technique that its designers will use to help build better automobile designs and concepts. Today’s designers already use publicly available text-to-image generative AI tools to assist in the earliest phases of design work. TRI’s new technique lets designers add initial design sketches and engineering constraints into the overall design process in the early stages, which helps reduce the number of iterations needed to create a sound engineering product.
Generative AI interprets patterns in existing data to create a wide variety of useful data for designers, including new and unique images, videos, audio, text, and 3D models. Capable of producing highly realistic and complex content, it has become a valuable tool in the gaming, entertainment, and product design industries. Several image-based AI programs are now commonly available as apps and cloud-based programs.
Toyota’s Step Forward
According to Toyota, generative AI has been useful as an inspiration to help stylists and designers but has yet to incorporate complex engineering and safety into its models. The new technique created by TRI combines Toyota’s formidable engineering expertise with a generative AI capability to produce a true design tool.
The principles used in optimization theory, a technique that is familiar to those who use computer-aided engineering, have been added to text-to-image-based generative AI. The algorithm allows the designer to optimize engineering performance while maintaining the text-based stylistic prompts applied to the generative AI process.
The designer can use text-based stylistic prompts to create an initial design concept. For example, a designer might request, with a text prompt, a series of designs based on an initial prototype sketch. The designer might provide additional input for a vehicle that is “SUV-like,” “sleek,” and “modern” and offer performance targets for the design. Then, by incorporating engineering models and factors like aerodynamic drag, ride height, interior and exterior dimensional constraints, and safety considerations into the AI technique, optimization can occur more quickly.
TRI’s generative AI technique optimizes aerodynamic drag in successive iterations based on parameter inputs from the designer. Image used courtesy of Toyota Research Institute
Designing the Next Generation Electric Vehicle
With the challenges in designing and building the next generation of EVs, the generative AI tool developed by TRI is expected to aid Toyota’s designers in its eventual move to electrification. It’s another tool to assist in building vehicles with optimized aerodynamic drag, safety, and interior comfort, that meet performance and range targets.
