Arm has launched the Total Design for Physical AI ecosystem, introducing the Robotics Capability Framework aimed at standardizing the description of robotic capabilities. Over 80 companies have joined this initiative.
Introducing Arm Total Design for Physical AI — bringing an initial 80+ companies together around a shared goal to accelerate physical AI from development to deployment.
One of the first initiatives is a Robotics Capability Framework, creating a common language for defining… pic.twitter.com/sBgpoT5aWw
— Arm (@Arm) September 8, 2026
From Reactive Robots to Self-Learning Systems
The Robotics Capability Framework aims to establish a standardized language for describing and comparing robotic systems. Arm proposes categorizing these systems into six complexity levels, ranging from RL0 to RL5, transitioning from reactive behaviors to contextual understanding, cognitive abilities, and self-learning.
Source: Arm.This framework does not seek to standardize specific hardware or software. Instead, it connects a robot's capabilities with real-world application scenarios and technical specifications, including latency, computation location, memory capacity, energy consumption, determinism, and safety.
Arm envisions that this will enable companies to provide clearer answers regarding what robots can actually do, rather than merely claiming "autonomy" or the use of physical AI.
Arm Gathers the Entire Stack
At the same time, the company has expanded its Arm Total Design program for physical AI, involving over 80 companies that operate across various technological layers. Participants include AWS, Hugging Face, NXP, QNX, Siemens, Unitree Robotics, Qwen, PSYONIC, Liquid AI, and others.
This initiative brings together software developers, AI model creators, sensor manufacturers, computing hardware providers, virtual platforms, and digital twin solutions. Arm believes that this approach will help reduce integration complexity and expedite the transition of robotics projects from prototypes to actual deployment.
For Arm, this is also a strategy to strengthen its position in the emerging physical AI market. The company is already promoting its architecture as the computational foundation for systems that need to perceive their environment, make decisions, and act within constraints of time, energy, and safety.
The Need for a Common Language in Robotics
Currently, developers can use the same terminology for vastly different systems, such as industrial robots, autonomous mobile platforms, or humanoid machines. This inconsistency makes it more challenging for clients to compare solutions and assess how well the claimed capabilities will perform beyond demonstrations.
Arm proposes addressing this issue through a unified description of capabilities and requirements while allowing developers the freedom to choose their own architecture, models, and software stacks. The company emphasizes that the framework should not standardize the performance, architecture, or implementation methods of robots.
Additionally, it was noted that in August, Wang Xiaogan, chairman of China's ACE Robotics, predicted a "ChatGPT moment" for robots by the end of 2027.