Rich Contact Sensing
Capture physical state where vision ends — force, slip, geometry, texture and motion at the point of interaction.
NOWTHE TOUCH INTERFACE FOR PHYSICAL AI
Intelligence becomes physical through touch.
We are building the interface that helps machines perceive contact, learn from interaction, and act with dexterity in the real world.
CONTACT
STATE
LEARNING
Sense, act, learn.Built for the physical world.
01 — THE MISSING MODALITY
Vision shows machines the world.
Touch reveals what is happening within it.
Most failures in manipulation happen at the point of contact — where objects slip, deform, resist, align or break. That state is often invisible. We make it observable, usable and learnable.
02 — FROM TOUCH TO INTELLIGENCE
A pathway from signal
to physical understanding.
Physical-State Interface
Translate tactile observations into synchronized, machine-usable signals for control, evaluation and recovery.
NEAR TERMLearning Interface
Connect touch, vision, proprioception, action and outcome — turning interaction into structured learning.
BUILDINGManipulation Intelligence
Create reusable representations and manipulation intelligence across tasks, tools and embodiments.
DIRECTION
03 — A PHYSICAL FOUNDATION
Designed from material
to machine intelligence.
Rich physical signals begin with a tightly integrated platform — materials, microstructures, optics, electronics, decoding and synchronization engineered as one system.
240K+SENSING POINTS / CM²
6-CHANNELPHYSICAL STATE3D displacement + 3D force
5M+NORMAL CYCLES100K+ SHEAR CYCLESdurability testing ongoing
<1 msMULTI-SENSOR SYNC*system-level
04 — WHAT BECOMES POSSIBLE
Machines that can
feel their way forward.
Touch is more than another sensor stream. It can become a new foundation for how robots adapt, recover and generalize.
Adaptive interaction
Manipulation that responds continuously to slip, friction, compliance and uncertainty.
Dexterity beyond vision
Robots that reason through occlusion, multi-contact and in-hand state.
Multimodal Robot Learning
Add synchronized contact state to demonstrations for training, evaluation, correction and recovery.
OUR DIRECTION
From tactile sensing
to manipulation intelligence.
Hardware is the beginning. The long-term opportunity is a physical intelligence layer that learns through contact and transfers across the machines that shape our world.
BUILD WITH US
What should machines
learn to touch next?
Tell us about the manipulation challenge, research direction or system you are building.