Analog Devices (ADI) presented the direction of next-generation robotics by unveiling 'Physical Intelligence (PI)' technology, which helps robots achieve human-like manipulation capabilities in real-world environments.

ADI participated in 'NVIDIA GTC 2026', held in San Jose, USA from the 16th to the 19th, and demonstrated robot manipulation technology combining sensing and artificial intelligence.

Physical intelligence is a technology that integrates sensors, signal processing, actuation, and AI to enable robots to perceive their physical environment and respond precisely.

At this event, ADI showcased key technologies that enable robots' sophisticated object manipulation capabilities, centering on an AI-assisted humanoid hand platform.

The platform was designed based on ADI's motor control and sensing architecture.

The operational precision of robots is still considered a challenge to be addressed in the automation industry.

Unlike humans, robots have shown limitations in tasks requiring tactile perception and fine force control.

Accordingly, ADI unveiled the first industrial robot precision benchmark to objectively evaluate advancements in robot manipulation technology.

This benchmark is physical-based sensing and highIt enables performance comparison and technology verification between robot platforms by utilizing fidelity simulation.

Three major demonstrations took place at the exhibition booth.

The first was a demonstration of 'tactile intelligence' using a multimodal tactile sensor, showing the process of a robot detecting contact and slippage without visual information and finding a socket along a cable.

This technology is designed to rapidly apply control methods learned in a digital twin environment to actual robots.

In the second demo, industrial robot sophistication benchmarks and a digital twin environment were introduced.

Technology that reduces the gap between simulation and reality, based on actual industrial tasks such as data center cable work or precision machine assembly, was emphasized.

This can reduce development costs and increase the reliability of robot operation algorithms.

Finally, an end-to-end manipulation demonstration was conducted in which an AI-based humanoid hand recognized the cable and connected it accurately.

This demo, integrated into a single system from sensing to control and execution, demonstrated the potential of robotics enabled by physical intelligence.

An ADI official stated, “Physical intelligence is a core technology that enables robots to work safely and precisely in real industrial environments,” adding, “It will accelerate the evolution of industrial automation by setting new standards for robot development and verification.”