Arm Holdings has announced the expansion of its Arm® Kleidi technology into the automotive market, aiming to enhance the development and implementation of automotive-specific Artificial Intelligence (AI) models. This decision reinforces Arm’s position as a strategic partner to leading tech companies in the sector, offering advanced solutions that drive performance and efficiency of AI applications in vehicles.
Suraj Gajendra, Arm’s Vice President of Automotive Products and Solutions, emphasized that while the industry mainly associates AI with the future of autonomous driving, the reality is that many functions in new models—such as adaptive cruise control, personalized infotainment systems, and monitoring of drivers and passengers—already rely on this technology and operate on the Arm platform due to its flexibility, performance, and scalability.
Unique Challenges for AI in the Automotive Sector
The advancement of these applications poses a challenge for the industry, which must face highly complex development environments, the need to deploy applications from the cloud to the vehicle, and comply with strict functional safety and cybersecurity regulations. Additionally, the longevity of vehicles requires that software be capable of updating and evolving over longer life cycles than other devices.
Arm addresses these challenges with its SOAFEE (Scalable Open Architecture for Embedded Edge) ecosystem, based on virtual platforms and standardized software solutions that accelerate the development and deployment of new functionalities.
Kleidi: Automatic Optimization with No Additional Effort
Launched in 2024, Arm Kleidi is an automatic optimization library for AI inference workloads on Arm CPUs. Without requiring additional tuning from developers, Kleidi is already integrated into leading frameworks such as PyTorch, MediaPipe, Llama.cpp, and ExecuTorch, improving the performance of AI applications both on mobile devices and in data centers, and now, also in automotive.
Its introduction to the automotive sector has already generated remarkable results:
- AWS Automotive has increased the response speed of its integrated vehicle chatbots tenfold, thanks to the integration of Kleidi into Llama.cpp. This allows drivers to inquire about information or resolve dashboard alerts without consulting manuals, reducing response times to just a few seconds.
- VicOne, specialized in automotive cybersecurity, has accelerated the response time of its xCarbon system, which detects security threats in real-time, improving token generation speed by 60% and reducing dependence on the cloud.
- Sonatus, with its AI Technician Builder platform for automotive OEMs, has migrated its workloads from GPU to Arm-based AWS Graviton EC2 instances, ensuring optimal performance and paving the way for executing AI in the vehicles’ edge computing.
94% of the Automotive Industry Trusts Arm
Currently, 94% of automotive manufacturers worldwide utilize Arm technology in their new models, reflecting the company’s strength in developing critical applications for connected and autonomous vehicles. The integration of Kleidi into this ecosystem offers developers tools that enable them to build high-performance AI models without additional investment, shortening development times and accelerating market access.
With this expansion into the automotive sector, Arm reaffirms its commitment to driving AI innovation from the cloud to the device, enabling companies to deploy advanced solutions that transform the driving experience and redefine the future of smart mobility.
via: Benchlife