Robots are no longer confined to factory workshops or science fiction movies – they have become an integral part of People’s Daily lives. They might be the silent sweeping robots working in the living room or the self-checkout machines guiding customers through shopping in supermarkets. These are no longer novelty but rather a microcosm of extensive technological changes.
Driven by artificial intelligence (AI) and machine learning (ML) technologies, robots are evolving from mechanical tools for performing tasks to collaborative partners with adaptive and decision-making capabilities. They can perceive the environment, respond in real time and even communicate through natural language. This increasingly enhanced intelligence is blurring the line between machines and assistants, opening up new possibilities for human-machine collaboration.
With the increasing maturity of technology, robots will play a greater role in more fields – from collaborative robots in factory workshops to autonomous systems in industries such as healthcare and logistics, and even to a wider range of application areas. AI not only makes robots more intelligent, but also makes them increasingly easy to collaborate with humans, possessing the capabilities of perception, decision-making and action, thereby enhancing efficiency and touching every aspect of our lives.
Arm is at the core of the evolution of physical AI
The world today is entering a brand-new stage of AI innovation. In this stage, intelligence no longer exists only in the cloud but is also embedded in real-world machines, endowing them with the ability to perceive the environment, make autonomous decisions and act in real time. This is the era of proxy AI (Agentic AI) : systems are no longer passively waiting for instructions but actively taking action, adapting in real time, and collaborating with other devices. Meanwhile, Physical AI is also on the rise – the same intelligence is embedded in physical devices, enabling them to have the ability of autonomous navigation, environmental perception and interaction with the real world.
Arm is at the core of this transformation. As a leading provider of high-efficiency and scalable computing platforms in the industry, Arm is promoting the deep integration of edge AI and robotics. In edge AI application scenarios, decisions need to be made immediately while also meeting the extremely low power consumption budget requirements. Whether it is the robots patrolling the station, the smart door locks that recognize faces, or the humanoid robots crossing the ruins, Arm’s computing architecture is the technical foundation that supports the realization of these “active intelligentization”.
The intelligent robot cluster and multi-machine autonomous collaboration system of R2C2
The AI platform of R2C2 is like the Android of the robot world – an interoperable operating system for heterogeneous robot clusters. In Hong Kong, the robot dog equipped with R2C2 can independently complete the inspection of train carriages, analyze thousands of visual data points, and the detection accuracy rate exceeds 99%. This system runs on the NVIDIA Jetson module with an Arm CPU. It processes low-latency operations such as control and detection through the CPU, while offloading high-load AI tasks to the GPU for processing.
R2C2 is tackling some of the most challenging problems in the field of robotics: interoperability among different types of robots, the scarcity of available application software, and the scalability of robot clusters. Centered on Arm technology, R2C2 is building a plug-and-play system, enabling it to operate stably and perform outstandingly in environments with limited connectivity, energy consumption and latency.
Cloud Deep Technology helps achieve the deep integration of sports and intelligence
Cloud Deep Technology’s quadruped robots are redefining mobility agility in complex environments – whether in industrial tunnels or post-disaster sites. This robot is built on the Rockchip RK3588 system-on-chip (SoC) equipped with the Arm architecture. While achieving high-performance motion control, its power consumption is just over 10 watts, which is one-third of that of traditional x86 systems.
This energy efficiency leap enables robots to be equipped with smaller batteries, more simplified cooling systems, and achieve longer battery life. Whether patrolling in power stations or evolving towards humanoid robot platforms, every step of CloudDeep Technology is driven by Arm cpus – providing precise and scalable computing capabilities for embodied intelligence in real-world scenarios.
Broadcom’s integrated AI chips empower edge intelligence
Broadcom’s BK7259 chip brings ultra-fast and ultra-efficient AI experiences to everyday devices, covering applications from toys to smart door locks. This chip adopts the Arm Cortex-M series core and the Eos-U65 AI accelerator. It can complete local face recognition in less than 200 milliseconds and has extremely low power consumption at the same time. It is an ideal choice for battery-powered edge AI devices.
Broadcom Integration has significantly accelerated the deployment and development process of AI applications by integrating Arm’s AI toolchain, including CMSIS-NN and Vela. Its platform supports real-time audio processing, has powerful security features and seamless Wi-Fi connection functions, fully demonstrating that powerful intelligence does not come at the expense of energy efficiency or cost.
The UCR humanoid robot achieves high cost-performance flexibility
Under Control Robotics (UCR) takes a different approach – creating robust, durable and cost-effective humanoid robots for high-risk working environments such as construction, mining and energy. Its prototype “Moby” adopts a lightweight AI model. This model has been trained in a simulation environment and deployed on the Jetson Orin module equipped with the Arm architecture, thereby achieving the capabilities of walking, balancing and environmental adaptation.
The system processes real-time motion logic through the Arm Cortex-A78 CPU and uses the STM32 controller to drive the motor to perform actions. With this configuration, UCR has achieved performance that was previously only available in high-cost systems, thanks to low-cost sensors and extremely low power consumption. This fully demonstrates how Arm empowers edge-native, AI-driven robots to operate efficiently in harsh environments – without relying on the bulky and power-consuming computing resources of traditional architectures.
Arm is the cornerstone of future physical AI
As the world enters the era of physical AI, robotics technology has transformed from a research field to an expandable practical application platform. From industrial quadruped robots to AI toys and then to humanoid robots, the common point of these diversified innovations lies in that they all rely on Arm’s heterogeneous computing architecture. By integrating energy efficiency, flexibility and scalable AI performance, Arm has built the underlying computing foundation for “proxy AI” and “entity AI”, making it not only possible from a technical perspective, but also having practical value for commercial application.
The future of robot technology will be autonomous, adaptable and deeply embedded. Whether it is smart cities, hazardous operation sites, or home service scenarios, Arm will always be at the core, providing powerful impetus for dynamic intelligence.
