Arm launches framework for physical AI and robotics
Arm has introduced Arm Total Design for Physical AI and a Robotics Capability Framework intended to create common technical foundations for automated systems. The initiative covers physical industries including mining, agriculture, manufacturing and transport, which together represent trillions of dollars in economic activity and an estimated annual compute opportunity of $200 billion by the 2030s. More than 80 organisations from the software, hardware and AI sectors are involved, including AWS, Hugging Face, NXP, Siemens and Unitree Robotics.
The framework addresses the absence of a shared way to describe and compare robotic capabilities. It defines progressive tiers ranging from reactive machines to context-aware, cognitive and self-improving systems. Each tier connects operational use cases with expected behaviours, outputs and hardware requirements, including latency, compute location, memory, power, determinism and safety.
Arm Total Design for Physical AI combines AI models, virtual platforms, digital twins, sensors, silicon and software for development before physical hardware is available. Arm demonstrated the approach with automotive partners using the Arm Zena CSS platform, where teams tested complex code before silicon production. Arm is seeking further technical contributions to expand the robotics framework.
Written by Sakhanda Wire AI from reporting by AI News. Facts, quotes and full context are in the original article below.