Note: The article was first published on TechNode China written by Icebin and translated by Zinan Zhang.
Car design, a task that has always been done manually — whether sketching, molding clay, or digital rendering — is gradually feeling the impact of AI and is set to undergo a fundamental transformation. Machine learning models can not only predict the appeal of new aesthetics but also generate attractive designs of their own. After being trained with large models, AI programs can run on standard enterprise laptops. AI models not only help designers work more efficiently but also generate new creative designs.
In the summer of 2000, General Motors introduced the Aztek, a crossover SUV that, despite having multiple features desired by outdoor enthusiasts, was heavily criticized for its appearance and often mocked as one of the ugliest cars ever made. The Aztek’s sales were only half that of the Buick Rendezvous, which was built on the same platform but redesigned and relaunched as the Buick Enclave, achieving success.
To avoid releasing the next Aztek, automakers are investing heavily in using AI to predict and generate attractive models.
Using AI to Predict and Generate Attractive Designs
Traditional theme clinics cost $100,000 a time, but with machine learning and deep neural networks, predictive models can significantly reduce the number of designs that need to be tested in these clinics. AI generative models create new car designs based on designers’ prompts, while predictive models can forecast consumer reactions to the designs.
As John R. Hauser, a professor of marketing at MIT Sloan School of Management, has said, the design of a car indeed has a significant impact on consumers’ purchasing decisions. With the advancement of AI technology, automakers can develop attractive models that meet market demand more efficiently and economically. By leveraging AI, automakers can maintain high design quality while significantly reducing design costs. For example, the design investment for a typical model may exceed $1 billion, and a major redesign could cost up to $3 billion. With AI assistance, these costs are expected to be greatly reduced.
AI has achieved a successful closed loop in automotive design
Data analysis and consumer preferences
AI can analyze vast consumer data, including purchase history, market trends, and social media feedback, to understand consumer preferences and needs. This data can help designers make more informed decisions when designing new models.
Generating design sketches
Deep learning-based image generation technologies (such as GANs) can produce numerous design sketches, helping designers explore different design options quickly. This allows the design team to gain more creativity and inspiration in the early stages.
Virtual Reality (VR) and Augmented Reality (AR)
Combined with VR and AR technologies, AI can create virtual car models for designers and customers to view and interact with in a virtual environment. This saves the cost of manufacturing physical prototypes and accelerates the design review and modification process.
Optimizing the design process
AI can optimize various stages of the design process, from material selection to aerodynamic analysis. AI algorithms can quickly simulate and evaluate the performance of different design options, helping designers make more efficient and accurate decisions.
User personalization
AI can also be used for personalized customization services, providing tailored design solutions and configuration options based on users’ specific needs and preferences. This not only improves user satisfaction but also enhances the brand’s competitiveness.
Toyota applies generative AI technology to automotive design
The Toyota Research Institute (TRI) has introduced generative AI technology to enhance the capabilities of automotive designers. This technology combines Toyota’s engineering strengths with the advanced functionalities of modern generative AI, allowing designers to reduce the number of iterations based on initial design sketches and engineering constraints. This technology not only inspires designers but also integrates practical engineering and safety considerations into the design process.
1. Incorporating engineering constraints into the design process
TRI’s new technology incorporates precise engineering constraints into the design process, such as aerodynamic drag and chassis dimensions. Drag impacts fuel efficiency, while chassis dimensions (like ride height and cabin size) affect handling, ergonomics, and safety. These engineering constraints can now be implicitly included in the generative AI process, enabling designers to optimize performance metrics while considering stylistic attributes.
2. Enhancing designers’ creative capabilities
The new generative AI technology allows designers to request a set of designs based on initial prototype sketches through text prompts, with specific stylistic attributes such as “stylish,” “SUV-like,” and “modern,” while also optimizing quantitative performance metrics. By reducing the number of iterations needed to align design and engineering considerations, this technology helps designers complete designs faster and more efficiently.
3. Improving electric vehicle design efficiency
By directly incorporating engineering constraints into the design process, TRI’s new tools help Toyota design electric vehicles faster and more efficiently. Takero Kato, President of Toyota’s BEV Factory, stated, “Reducing drag is crucial for improving the aerodynamic performance of BEVs and maximizing their range.” This technology enhances the range and overall design efficiency of electric vehicles by optimizing aerodynamic performance.
GAC’s breakthroughs and innovations in AI automotive design
1. AI-driven design optimization
GAC Group has made significant breakthroughs in AI automotive design, particularly in design optimization. By utilizing artificial intelligence technology, GAC can quickly generate and evaluate multiple design options. This not only accelerates the design process but also ensures that each design achieves optimal aesthetics and functionality. AI models can analyze vast amounts of data to predict market trends and consumer preferences, helping designers make more informed decisions.
2. Intelligent design assistance system
GAC has developed an intelligent design assistance system that combines machine learning and computer vision technology. Designers need only input basic parameters and design requirements, and the system can generate multiple design sketches that meet these criteria. These sketches not only include exterior designs but also consider factors such as aerodynamics, material usage, and manufacturing costs. This intelligent assistance system greatly improves design efficiency and quality.
3. Application of virtual reality and augmented reality technologies
To further enhance the design experience, GAC has also integrated virtual reality (VR) and augmented reality (AR) technologies into the design process. Designers can view and modify car designs in a virtual environment, and customers can preview customized models through AR technology. This immersive experience not only enhances the intuitiveness of the design but also speeds up the design review and modification process, giving GAC an advantage in a highly competitive market.
In conclusion, the integration of AI in car design is ushering in a new era of innovation and efficiency. By leveraging machine learning models, automakers can predict consumer preferences and generate aesthetically pleasing designs more quickly and cost-effectively than traditional methods. These advancements not only enhance creative capabilities but also optimize design efficiency, ensuring that new models meet both aesthetic and functional standards. As AI continues to evolve, its role in automotive design will likely expand, enabling automakers to produce cutting-edge vehicles that resonate with consumers and maintain a competitive edge in the market.
