[ad_1]
Recent advancements in digital twins have enabled automakers to implement AI-driven solutions, enhancing both efficiency and innovation. By Pedro Moreno
Thirty years ago, the automotive industry experienced a pivotal change with the integration of electronics into vehicles, necessitating new skills and working methods. The introduction of active safety features, automated driving capabilities, and digital interfaces further accelerated this transformation. The emergence of over-the-air (OTA) updates marked the beginning of a new phase, allowing vehicles to adapt post-release with additional features and services for which customers were willing to pay. This marks the era of the software-defined vehicle, where software functions are largely performed within the vehicle itself.
The landscape shifted again with OpenAI’s launch of ChatGPT in 2023, prompting various industries to recognize the transformative potential of generative AI. Capgemini predicts that approximately 16% of major automotive original equipment manufacturers (OEMs) stand to increase their operating profits by scaling up AI applications.
AI is set to enhance not only vehicle functionalities but also improve efficiencies throughout automotive production workflows. This encompasses a range of activities, from designers employing AI-generated images to envision new vehicles, to engineering teams using generative AI APIs to create digital twins of their facilities, as well as marketing departments leveraging AI tools to brainstorm marketing content and advertising strategies.
Most automakers are currently assessing between 100 to 150 potential AI use cases, making it difficult to conceive of any function that won’t benefit from generative AI. This shift heralds the arrival of the “AI-defined automaker.”
Knowledge Chatbots Boosting Customer Service
Ford, which fields around 15,000 customer calls daily, has integrated generative AI into its call center operations. A speech-to-text system initially processes customer inquiries, stripping out personally identifiable information in line with compliance regulations in various countries. This efficient data management paves the way for a secondary AI service—a chatbot that aids supervisors and engineers in retrieving data, such as the volume of calls or specific recurring issues. Moreover, AI supports call center operators by suggesting optimal responses to customer inquiries, thereby enhancing the speed, efficiency, and overall satisfaction in customer service.
In addition, knowledge chatbots are utilized in repair shops and dealerships. For instance, when a vehicle’s air conditioning system fails, technicians can simply inquire about the optimal repair procedure for that specific model, and the chatbot will present detailed instructions along with visual aids.
Automated Vehicles 2.0
The earlier approach to developing automated driving functionalities, termed AV 1.0, relied on modular systems focusing on perception, prediction, and actuation. The perception component predominantly utilized Convolutional Neural Network (CNN) classification AI. It was widely believed that achieving perfect object classification in driving contexts would solve the problem of autonomous driving at higher levels of automation. However, after more than a decade of efforts, it became evident that the vast array of driving scenarios could not be comprehensively addressed through perception alone; an accurate scene interpretation is essential for managing unfamiliar situations.
Conversely, the transformer engines that power generative AI are trained on broad datasets, not just automotive information, allowing them to abstractly comprehend and make sound decisions in previously unseen driving conditions. This marks the foundation of AV 2.0, wherein an end-to-end system utilizes raw sensor data as input and directs vehicle actions—acceleration and steering—as outputs, superseding the traditional modular architecture of AV 1.0. This method not only enhances performance but also optimizes the allocation of computational resources by reducing competition among various function modules. It also simplifies the transfer of vehicles across different urban environments, a challenging task for rule-based AV 1.0 systems.
Startups such as Wayve, focusing on embodied AI for assisted and automated driving, and Waabi, working on next-generation self-driving technologies, have made remarkable strides in this domain.
The Power of Digital Twins
Digital twins are accurate virtual representations of physical assets, processes, or environments. In industrial applications, digital twins are invaluable throughout the entire product lifecycle—from design and manufacturing to maintenance, operations, and marketing.
These digital counterparts facilitate realistic visualization and precise simulation of various product iterations and scenarios. They empower teams to explore virtual factories, examine vehicle prototypes, and understand past mistakes to foresee future challenges. Instead of relying on face-to-face meetings and static planning documents for project alignment, digital twins foster streamlined communication, ensuring that vital design choices are driven by up-to-date information. Stakeholders can visualize designs in situ, enabling early error identification and feedback integration during the review process.
Recent advancements in digital twin technology include the use of OpenUSD (Universal Scene Description), an adaptable ecosystem designed for the description, composition, simulation, and navigation of 3D scenes. Just as HTML serves the 2D web, USD acts as a powerful, extensible, open language for the 3D web, fostering communication across various tools and departments involved in the product lifecycle. Innovations in physics-based rendering and scalable computing have made it possible to create large-scale, realistic digital twins that accurately reflect the physical world.
In early 2022, BMW launched the iFACTORY initiative for virtually planning using digital twins. The BMW Group conducted 3D scans of its vehicle and engine manufacturing plants, encompassing over seven million square meters of indoor space and 15 million square meters of outdoor facilities. The company employs a significant team of factory planners worldwide, including those engaged in the projected Debrecen facility, whose roles are complex and critical; minor miscalculations can lead to substantial financial repercussions.
Thus, the value of precise virtual planning cannot be overstated as it allows planners to experiment and implement changes with minimal cost. They can optimize production processes in virtual environments before committing to large-scale construction projects and capital investments, significantly reducing expenses and downtime associated with change orders and flow re-optimizations in existing factories.
The synergy of digital twins and generative AI further amplifies the benefits of each technology. Generative AI can create digital assets, enabling teams and customers to interact with their knowledge bases using natural language, navigate digital twin environments, and develop visually and physically accurate digital realms for purposes like training autonomous vehicles or robots. The digital twin thus becomes a testing ground for developing the software functions of products. In addition, digital twins are utilized for generating synthetic data, which is essential for training the AI algorithms associated with the developing product.
Welcome to the Era of AI-Defined Automakers
The automotive sector is experiencing a significant transformation, moving beyond just introducing software-defined vehicles to enter the realm of AI-defined automakers. Generative AI is reshaping every facet of the industry, including design, engineering, customer service, and autonomous driving. With the aid of digital twins’ advancements, automotive manufacturers are swiftly implementing AI solutions aimed at enhancing efficiency, fostering innovation, and improving customer experiences. As the industry leans into this AI-centric future, every aspect stands to gain and be optimized through generative AI, paving the way for greater efficiency, creativity, and a focus on the customer, setting the stage for ongoing success in the automotive landscape.
About the Author: Pedro Moreno Lahore is a Senior Account Executive in the Automotive sector at Nvidia.
.
[ad_2]
