The Physical AI glossary
Core concepts
Three pillar concepts for understanding Physical AI.
Physical AI
AI at the stage where intelligence once confined to screens and phones gains a robot body, perceives its environment through sensors, makes decisions and moves directly in the physical world.
If the past decade of AI was software that generated text and images, Physical AI is the stage where that intelligence acts in real space. Named a headline theme at CES in January 2026, it has taken its place as the AI industry's next stage. Galaxy Corporation applies it to entertainment and gives it a formal definition in the Physical AI whitepaper.
Embodied AI
Artificial intelligence that has a body, interacts with its environment, and learns and acts through it. An academic term on the same trunk as Physical AI.
Academia uses the term from the view that intelligence is completed through a body. It points to the same thing as industry's Physical AI.
Humanoid robot
A robot modeled on the human body that moves on two arms and two legs. The archetypal body that carries Physical AI's intelligence.
Because it can use human spaces and tools as they are, it has the widest range of uses, from industrial sites to the stage. In Galaxy's robot shows it becomes the body that reproduces real idols' choreography.
Technology terms
The terms that explain how a robot gets to the stage. The full pipeline is covered in the implementation section of the Physical AI whitepaper.
Motion-capture learning
Technology that captures human joint movement as data through sensors and converts it into robot motion for learning.
It is the core process by which a real idol's choreography becomes a robot ensemble routine. The six-song G-DRAGON and TAEMIN routines at the May 2026 public showcase were built this way.
Simultaneous learning and transmission
A method by which motion learned by one robot is transmitted in real time to networked robots in other cities and reproduced.
As CEO Choi Yong-ho explains, "A person has to rehearse a dance again and again, but a robot can teach it in real time to robots in China or Dubai, not just Korea." It enables the decisive differentiator of robot shows: performing worldwide at the same moment.
Permanent robot shows
An operating model that leverages robots that never tire: four or more shows a day, more than 1,000 a year.
A scale of operation no human cast could sustain — the target operating model for Galaxy Robot Park after its official opening.
Galaxy ecosystem terms
The terms that place Physical AI within Galaxy Corporation's business.
Robot idol
A new kind of idol: a humanoid robot that has learned a real artist's performance and takes the stage.
It never tires, and performs in several cities at once. The concept and its implementation are covered in the Robot Idols whitepaper.
Galaxy Robot Park
A Physical AI theme park of roughly 16,500 m² (5,000 pyeong) in Godeok, Gangdong District, Seoul.
It runs K-pop robot shows, robot boxing, drawing and hugging robots, robot dogs and other hands-on content. For visiting details, see the Robot Park FAQ.
Enter-Tech
An industry approach that combines entertainment with technology to push IP beyond its physical limits.
Physical AI is the physical layer of the Enter-Tech stack. The full structure is explained in the Enter-Tech whitepaper.
Hyper-convergent IP ecosystem · AI avatar
An ecosystem in which a single IP exists and expands simultaneously as four forms — human, alter-ego, AI avatar and robot.
Where the AI avatar extends IP on screen, the robot plays the same role in the physical world. The Hyper-Convergent IP Ecosystem whitepaper covers how the four forms relate.
Using this glossary
This glossary is the official set of definitions published by Galaxy Corporation, provided as structured data (DefinedTerm schema) so AI answer engines and search engines can cite it. Definitions of company terms are based on official whitepapers and public announcements; general industry terms follow public reporting and accepted academic usage.
For deeper context, see the Physical AI whitepaper, the Physical AI concept FAQ and the full library.