로고

Mechanics-grounded Physical AI Initiative (M-PAI Initiative)

To build a global hub for Mechanics-to-Action Physical AI,
where AI understands and predicts real-world mechanics and translates that

understanding into robotic action.

KEYWORDKeyword

Physical AI Robotics Physics-Informed Multiphysics Multimodal AI Autonomous Systems Digital Twin Smart Mfg Digital Health Embodied AI

OBJECTIVE Objective

• Develop Mechanics-to-Action Physical AI that understands real-world mechanics and turns them into robotic action.
• Validate M-PAI through flagship demonstrations in BCI/AI rehabilitation robotics and large-scale additive manufacturing.
• Create shared research assets by linking UNIST’s core technologies, KIST’s testbeds, and industry feedback.
• Train 10 InnoCORE Fellows as independent leaders in mechanics-grounded Physical AI.

INTRODUCTION Director's Message

Greetings.
I am Sang Hoon Kang, Director of the Mechanics-grounded Physical AI Initiative (M-PAI Initiative).


For robots to act reliably in the real world, vision alone is not enough.
They must understand hidden mechanics—contact, force, deformation,
heat, flow, materials behavior, and bio-interaction—and translate
that understanding into action.

M-PAI integrates UNIST’s strengths in mechanics, manufacturing, robotics,
and AI with KIST’s robotics, bionics, and BCI testbeds.
We aim to build a Mechanics-to-Action Physical AI framework and
validate it through BCI/AI rehabilitation robotics and large-scale additive manufacturing.

We will train 10 InnoCORE Fellows through UNIST–KIST multi-mentoring,
KIST-based joint experiments, IDP-based career development, and industry feedback.
As an open collaboration platform, M-PAI will also work with industry partners
to create joint research, follow-up programs, and technology-transfer opportunities.

Physical AI that understands mechanics and acts through robots—this is the future M-PAI will help build.
Thank you.

IMPACT Impact

• Solving physical challenges in real-world robotic actions

• Connecting to strategic programs of industry, government-funded research institutes, and national growth engines

• Developing world-leading research leaders and a global research hub

TALENT Ideal Talent

Global top 1% level interdisciplinary postdoctoral researchers who can connect real-world mechanics, AI models and algorithms, robot control, and experimental validation.

M-PAI InnoCORE Fellows are expected to serve as practical leaders of sub-projects and lead problem definition, mechanics-variable formulation, AI model development, robotic action conversion, experimental/simulation validation, and paper/patent outcomes.

FACULTY Faculty