About Us
A robot simulation and physical scoring company for Physical AI
Mission
We want to make it routine to check whether a robot or AI can really do the customer's job before it goes into the field.
Vision
Through scoring that discloses its reasons and limits, we are building workplaces where robots and people can work together with confidence. Our scoring method (PAVS) is public, and our own system is checked against the same criteria.
Core Values
Physical scoring
We judge by measured contact, position and force — not by a "success" log.
Three-state verdict
Results are pass, fail or not measured, and anything not measured never counts as a pass.
Self-verified scorer
Negative controls and regular automated checks test the scoring program itself.
Stated limits
Measurement scope, data sources and the stage of real-robot comparison are written into every result.
Contact
Headquarters
#315, LH Business Growth Center, Pangyo 2nd Techno Valley 54 Changeop-ro, Sujeong-gu, Seongnam-si, Gyeonggi-do, Korea
Weekdays 09:00 - 18:00 (KST)
Our Journey
Our company's key milestones
Founding and Cloud XR
- Company founded in 2017
- Developed cloud XR streaming technology
- Cloud XR commercialization partnership with Deutsche Telekom and SKT (2021)
Digital Twin
- Four years of government-funded digital twin research
- Commercialized a 3D digital twin solution (2025)
imSim Robot Simulation
- imSim MVP development completed (January 2026)
- Test environments built for robot arms, mobile robots, and defense unmanned systems
AI Robot Adaptation & Validation
- Released PAVS, an open specification for scoring AI robots (September 2026)
- Released the revised PAVS (October 2026)
- Joint pilot tests and expanded real-robot comparisons (planned)
Our Vision
Making it standard practice to check whether robots and AI can do the job before they go into the field
Adapt to the task
Further training on test records, then retesting with the same scenario
Score physically
Pass or fail for each item, judged from contact, position and force measurements
Open about limits
Verdict reasons, measurement scope, and limits are recorded alongside the results
Current stage and resolution conditions
For each item, we list its current status alongside the condition for resolving it. Our progress is this list getting shorter.
| Item | Current status | Resolution condition |
|---|---|---|
| Match with real robots | We compared against a real robot once (2026-09-02) but judged the evidence insufficient. Whether task scores match real robots has not yet been confirmed. | Comparative measurements against real robots along multiple axes |
| Image detection domain | In crack detection, the detection ranking by crack-width band matched the ranking from real measurements (2026-08). Only the ranking matched, not the level — the simulation is more optimistic. | Re-measurement and extension to other domains |
| Results report issuance | We generate reports, and a report is not issued if it lacks the measurement scope and an adaptation-involvement statement. A product for external issuance is not yet ready. | Connect tamper detection and define the product |
| Official status | We are not a certification or accreditation body, nor do we have any external recognition. Standard numbers are for reference only. | Once the institutional framework is in place |
| Scoring external robots | All scoring to date has been on our own subjects (in-house simulation and our own real robots). We have not yet scored any customer-commissioned robot. | Customer-commissioned tests |
| Business stage | The adaptation workflow is in place, but it has not yet been offered as a paid service. | First paid adaptation contract |
| Differences by robot type | Cross-checking whether a task “really worked” is being applied first to robot-arm pick-and-place tests; other types are not yet at the same level. | Physical evidence defined per type |
Three levels of simulation trust
These three levels are our only grounds for claiming a match with reality (as of October 2026).
Press Coverage
Press coverage archive
IMMERSIVECAST Releases 'PAVS', an Open Specification for Scoring AI Robots
Released 'PAVS', an open specification defining how to score · validate AI that moves in the physical world in simulation · white paper · specification text · case book · own conformance statement published together on the website
IMMERSIVECAST Completes 'imSim' MVP Based on Digital Twin Research
Completed MVP of AI-based robot simulation platform 'imSim' built on 4 years of digital twin research. Participating in government last-mile delivery robot R&D project, preparing for full market launch in 2026.
IMMERSIVECAST Accelerates Commercialization of 3D Digital Twin Solution 'VRaum Twin'
After completing 4 years of government-funded research, launching commercialization of a digital twin solution that recreates manufacturing facilities, robots, and worker movements in real-time 3D virtual space.
End of Low-Quality 3D Era... 8K Content Without High-End PC, How?
Developed cloud XR technology that streams 8K+ high-resolution content at 90fps or higher regardless of device specifications.
IMMERSIVECAST Partners with Deutsche Telekom and SKT for Cloud XR Commercialization
Collaborating with Deutsche Telekom and SK Telecom to develop commercial cloud XR services including virtual car showrooms and industrial training simulations.
We are hiring
Join us in building robot and AI scoring and adaptation — careers inquiries
contact@immersivecast.com