Instant Physical AI upgrade
Turns passive ground and aerial platforms into self‑navigating autonomous agents without system redesign.
PAI‑X: Physical AI Innovation Startup
Any platform becomes an autonomous agent.
Available Dec 2026
PAI‑Lynx
PAI‑Lynx turns passive mobility and robots into autonomous Physical AI. The system builds an N‑dimensional fusion map in real‑time, providing rich information enough to give any mobility or robot full autonomy capability in any operating condition.
View productCompany
A deep‑tech company building intelligence that acts in the physical world.
PAI‑X builds the bridge between basic robotics and true Physical AI. Our modules turn standard mobility platforms into mission‑ready autonomous systems.
What the name means.
- PAI
- The connection between artificial intelligence and the physical world.
- Ø
- The mark itself: the spirit of creating 有 out of 無.
- 파이
- Said aloud in Korean, PAI is 파이, a pie. Technology made to nourish the people who use it.
- X
- Fusing modalities normally kept apart, and expanding the reach that gives a machine.
Physical AI
From nothing, something
Pie
Cross & eXpansion
At a glance.
- Company
- PAI‑X
- Representative
-
Ki‑Yong Oh
Associate Professor, Hanyang University - Founded
- August 2026
- Headquarters
-
222 Wangsimni‑ro, Seongdong‑gu, Seoul
Hanyang University - Field
- Robotics, Vehicle Tech & Advanced Mobility
- Business model
- B2B · B2G · Subscription
- Core technology
- Quality‑adaptive fusion of four sensing modalities
- First release
- PAI‑Lynx, 1 December 2026
How we get there.
-
STEP 1
Public infrastructure & industry
Enter the market for autonomous inspection of power and fire‑safety installations.
-
STEP 2
Robotics
Supply the perception module to indoor and outdoor autonomous mobile robots.
-
STEP 3
Global platforms
Expand into self‑driving and self‑flying vehicles and UAV autonomy platforms.
Product
The line‑up.
-
PAI‑Lynx
Turns passive mobility and robots into autonomous Physical AI, building an N‑dimensional fusion map in real‑time.
Available Dec 2026
PAI‑Lynx
Full sensor suite, high‑performance edge computer and dynamic fusion software, consolidated in one compact all‑in‑one housing.
- 4
- Modalities fused as equals:
3D LiDAR, IMU, optical, thermal - 100%
- Onboard sensing, fusion and
mapping, all run at the edge - 0
- Dominant‑sensor dependencies.
Zero‑modality‑failure architecture - Dec2026
- First commercial availability
Hardware
- Rigid sensor geometry
- Sensors locked in fixed relative position so calibration survives rough terrain and high vibration.
- Motorised tilt head
- Angles the sensor head dynamically to widen the field of view and remove upper blind spots.
- Light on the payload budget
- One housing replaces a stack of separate sensor payloads, so little of the platform's carrying capacity is spent on perception.
- Zero‑expertise deployment
- Calibration, quality scoring and stream reweighting all run automatically onboard.
Reliability decided live, inside the estimator.
-
01 ALIGN
One spatial frame
Custom sensor geometry, hardware‑level time synchronisation and dynamic extrinsic calibration bring every stream into a single frame, so geometry, appearance, motion and heat all describe the identical instant.
-
02 WEIGH
Authority that moves
Reliability is scored continuously from geometric consistency and feature richness, inside the tightly coupled estimator rather than as pre‑filtering. Enter smoke and optical weight drops while thermal and LiDAR take precedence.
-
03 ACT
One trusted output
The platform receives a single ROS‑compatible spatial output straight into its motion controller. No external compute, no sensor expertise, no platform redesign.
Capabilities
Unified N‑dimensional mapping
3D metric geometry, visual appearance, thermal signature and vehicle pose combined into one actionable frame.
Uninterrupted mission resilience
Zero‑modality‑failure architecture holds spatial density even when darkness, smoke or dust disables individual sensors.
GNSS‑denied navigation
Continuous relative localisation indoors, underground and in jammed environments where satellite signals fail.
3D spatial thermal recognition
Maps heat anomalies such as fires, overheating assets and hidden personnel directly in 3D space rather than flat 2D imagery.
100% onboard edge compute
Sensing, dynamic fusion and mapping all processed locally, for zero‑latency intelligence with no cloud dependency.
Where it is used
Critical infrastructure inspection
Unmanned vehicles inspect tunnels, ports and power grids autonomously, with no human operator in the loop.
Logistics & manufacturing
Conventional material‑handling platforms become self‑operating Physical AI units, cutting labour cost and raising throughput.
Zero‑visibility missions
Resilient autonomous operation in smoke, darkness and GPS‑denied environments for defence and emergency response teams.
Contact
Put Physical AI on your platform.
Tell us what you operate and the conditions it has to survive. We will walk you through integration and the fusion output your motion controller receives.
- munsu3744@hanyang.ac.kr
- Phone
- +82‑10‑3275‑3744
- Address
-
Room 906, Mechanical Engineering Building
Hanyang University, 222 Wangsimni‑ro
Seongdong‑gu, Seoul, Republic of Korea