Global automakers are under tremendous pressure to innovate more quickly as a result of the most rapid engineering changes that cars have seen in their 100-year history.
This includes Toyota Motor Corporation, the world’s largest automaker with over 10 million vehicles sold last year.
Even as thousands of engineers retire, Toyota is creating a system of generative AI agents to store and exchange internal knowledge in order to create new car models more quickly.
The automotive engineer in charge of the generative AI project, Kenji Onishi, says the company is changing from being a car manufacturer to a mobility company.
“The largest obstacle is the rapidly growing number of items that need to be developed.” “Kenji Onishi said.”
The shiny products that come out of Toyota factories all over the world now include a variety of hardware and software, such as batteries and charging stations.
“O-Beya” is the system’s name, derived from a long-standing Toyota management term for cooperative teams. “O-Beya” translates to “large room” in Japanese. The goal is to create a large room full of AI agents, or experts, who are available around-the-clock by using design data from actual engineers.
There are currently nine AI agents in the “O-Beya” system, ranging from a vibration agent to a fuel consumption agent. Multiple agents can be chosen by users to respond to a query.
An engineer might ask O-Beya, for instance, how to improve a car’s performance. O-Beya combines the responses of a regulatory agent regarding emission limits and an engine agent regarding engine output into a single response.
In the future, users won’t have to choose the appropriate agents because the system will do it for them.
The division that creates powertrains—the essential component that joins the engine and wheels to propel a vehicle—uses O-Beya.
Powertrain design necessitates the collaboration of numerous specialists in fields such as engines, batteries, driving, and even sound.
According to Onishi, these specialists are comparatively old, and their expertise will be lost after their retirement. The goal here is to stop it from happening, he stated. Therefore, Toyota wish to impart this knowledge to the next generation for this reason.
Toyota’s AI system is built on Microsoft Azure OpenAI Service and uses OpenAI’s multi-modal GPT-4o Large Language Model (LLM).
Azure Functions, an API or application programming interface, connects Azure OpenAI Service with Azure Cosmos DB, an AI-ready database that enables vector search – a type of search that can find closely related information beyond just keywords.
The proprietary system is grounded with Toyota’s design data that includes its past engineering design reports, the latest regulatory information and even handwritten documents by veteran engineers.
In order to continuously improve, Toyota can also safely store user conversation histories and human expert reviews of the AI responses using Azure Cosmos DB.
This Is The 2024 Toyota Corolla Cross Hybrid Nightshade
According to Onishi, the data set will eventually contain non-text information such as technical drawings.
Since January 2024, O-Beya has been available to about 800 engineers who work on powertrains, which include the engine, transmission, driveshaft, axles, and more. According to Onishi, it has been used “hundreds of times” every month.
“It’s much easier to find information on Toyota’s AI..,” as per Onishi.
By volume, Toyota is among the largest automakers in the world. According to the Japan Times, it achieved a record net profit of $31.9 billion for the fiscal year that concluded on March 31, 2024. Toyota also produces high-end vehicles under the Lexus name.
The number of AI agents increases along with the complexity of car manufacturing.
