BREAKING  Summer Robotics closes Series A led by Applied Ventures Kortx sees metal, glass + paint - the surfaces cameras hate 20ms reaction time / 100µm precision / 100Hz tracking From string theory to robot vision 4 event sensors. 12 lasers. Zero cameras. BREAKING  Summer Robotics closes Series A led by Applied Ventures Kortx sees metal, glass + paint - the surfaces cameras hate 20ms reaction time / 100µm precision / 100Hz tracking From string theory to robot vision 4 event sensors. 12 lasers. Zero cameras.
Profile / Robotics & Perception

Schuyler
Cullen.

He studied extra dimensions in graduate school. Now he builds the eyes that let robots find the edge of a pane of glass.

Co-Founder & CEO Summer Robotics Campbell, CA
On the record Schuyler Cullen, co-founder and CEO of Summer Robotics

Schuyler Cullen. A theoretical physicist who decided the most interesting unsolved problem wasn't in the universe - it was in the factory.

What he is building now

The camera was the problem. So he threw it out.

Schuyler Cullen runs Summer Robotics, a company that decided robots would never really see the world as long as they were looking at it through a camera.

Most machine vision works the way you'd expect: a camera takes a picture, software guesses what's in it. That works fine on a printed label under good light. It falls apart on a chrome bumper, a wet glass windshield, or a glossy painted door - the exact surfaces that fill a real factory. Reflections lie. Highlights blow out. Transparent parts simply vanish. Cullen's bet, made with co-founder Dirk Smits, was that you can't patch your way past that. You have to rebuild perception from the photon up.

Their answer is a platform called Kortx. Instead of a camera, it uses lasers and event sensors - chips that don't record frames but fire the instant a point of light changes. The result is a continuous 3D stream rather than a stack of still pictures. It tracks at 100Hz, measures to roughly 100 microns - about the width of a human hair - and reacts in 20 milliseconds, five to ten times faster than camera-based systems. It can be set up on a line in about half an hour. And it sees the metal, glossy, and transparent surfaces that blind everything else.

The point isn't speed for its own sake. It's what speed buys you on the floor: jig-free assembly, multi-SKU production, robots that adapt instead of waiting for the world to hold still. In 2025 the company closed its Series A, led by Applied Ventures - the venture arm of Applied Materials - with Solasta Ventures and NAVER D2SF joining, and began running Kortx pilots with automotive manufacturers.

Kortx, by the numbers

Numbers that read like a dare

100µm
Measurement precision
20ms
Reaction time
100Hz
Tracking rate
5-10×
Faster than cameras
Reaction speed vs. camera vision5-10x
Surfaces handled: metal / glossy / transparentfull set
Setup time on a line~30 min

"Our Kortx platform is enabling robotics applications that were not feasible with existing solutions."

- Schuyler Cullen, on the Summer Robotics Series A, 2025
How Kortx actually sees

Eyes, not snapshots

Seeing like an animal, not a camera

4 EVENT SENSORS · 12 LASERS · ONE CONTINUOUS 3D STREAM
STEP 01
Recognize
Pick out shapes and materials - including the transparent and glossy ones cameras miss.
STEP 02
Track
Follow motion at 100Hz, locking onto metal and glossy surfaces in real time.
STEP 03
Measure
Resolve 3D surface detail to roughly 100 microns - the width of a hair.
STEP 04
Control
Feed the robot fast enough to react and avoid collisions in 20 milliseconds.
Camera vision
  • Frames, then guesses
  • Blinded by glare + glass
  • Needs controlled lighting
  • Slower to react
Laser-event sensing
  • Continuous 3D stream
  • Reads metal, gloss + transparent
  • Works across lighting
  • 20ms reaction time
The long way around

A string theorist walks into a factory

Before the lasers, there was light of a different kind. Cullen did his undergraduate work at Caltech, taking degrees in both mathematics and physics, then went to Stanford for a PhD in theoretical physics. His doctoral world was string theory and particle physics - the kind of work that produces papers with titles like "TeV Strings and Collider Probes of Large Extra Dimensions," a 2000 study of how you might catch a glimpse of hidden dimensions inside a particle collider.

It's a strange apprenticeship for a robotics CEO, and also a perfect one. String theory is, at bottom, a discipline about how light and matter behave when you push them to the edge of what's measurable. Two decades later, that's still the job - only now the measurement happens on an assembly line instead of in an accelerator.

The bridge from theory to product ran through video. Cullen worked as a scientist and project leader at Pulsent on image processing, then co-founded and ran Keystream, a startup that built algorithms to automatically track, extract, and recognize objects in video - and pointed that technology at web advertising. Teaching software to find an object inside a moving image is, it turns out, a long warm-up for teaching a robot to find a part inside a moving cell.

Then came Samsung. As Senior Director and later Vice President of AI and Robotics, Cullen led the teams building the company's autonomous-mobility stack - the algorithmic brain for robots and self-driving cars - and ran research partnerships with Berkeley's BDD and BAIR labs. He also worked the investment side, backing AI-processor, sensor, and software companies. He'd seen, from the inside, what it takes to get a machine to move through the physical world. He'd also seen where it kept getting stuck.

It kept getting stuck on perception. Not on the muscles, not on the planning - on the eyes. So in 2020 he left to fix the eyes, and Summer Robotics was born.

"Robots don't have a muscle problem. They have an eyes problem."

- The thesis behind Summer Robotics, in plain terms

Why the small numbers are the big story

100 microns is the whole pitch

A hundred microns is roughly the width of a human hair. It is also the difference between a robot that can do precision assembly and one that can only do choreography.

The robotics industry spent years selling the body - arms, grippers, humanoids that walk. Cullen's company is selling the sense the body needs to be useful. A robot that can only act on a part it knew about in advance, in a fixture that holds it perfectly still, isn't really adapting to the world. It's performing a memorized routine and hoping nothing moves.

Kortx is built for the opposite case: parts that aren't where you left them, lines that switch between products, surfaces that reflect and refract. That's why the market language around Summer Robotics is all about jig-free assembly, multi-SKU production, and adaptive robot control. Take away the fixtures and the single-product line, and you take away most of the cost and rigidity of factory automation. The 20-millisecond reaction time is what makes a robot safe enough to share that space - fast enough to see a collision coming and stop.

It's a deeply unglamorous problem dressed up in deeply elegant physics. Which, if you've followed Cullen's path from extra dimensions to dented car doors, is exactly on brand.

Things you can drop at a dinner party

Five facts worth keeping

01

He co-authored a particle-physics paper on detecting large extra dimensions at colliders - then pivoted to robots.

02

Kortx runs on 4 event sensors and 12 lasers. There is no traditional camera anywhere in the system.

03

The system is designed to see transparent and glossy parts - the exact materials that defeat conventional machine vision.

04

Two of the hardest schools in the country back-to-back: Caltech for undergrad, Stanford for the PhD.

05

Before robots, he taught software to recognize objects in video ads at his startup Keystream.

The shareable version

If this profile were a headline

From string theory at Stanford to teaching robots how to see. The thread that connects them: light.
No cameras. Just lasers and event sensors - animal-like vision at 100Hz.
The surfaces that blind every camera are exactly what his platform sees best.
Why teach a robot to see like a camera when you can teach it to see like an eye?
He built Samsung's self-driving brain. Now he's building eyes for the robots that replace the jig.
A startup that scrapped the camera and rebuilt perception from the photon up.
Follow the work

Where to go next