The company teaching robots, cars and checkout counters to read the third dimension.
FILE PHOTO: An Aurora 930 structured-light camera, roughly the size of a deck of cards, staring back at a world it measures in millimeters. It has no idea it is replacing a LiDAR puck ten times its price.
Somewhere right now, a warehouse robot turns a corner it cannot afford to misjudge. A car decides whether the person in the driver's seat is awake. A hand hovers over a payment pad and the machine asks a quiet question: is this a real palm, or a photograph of one?
Behind each of those moments there is a decent chance a Deptrum sensor is doing the looking. The company makes depth cameras - hardware that does not just photograph the world but measures it, point by point, in three dimensions. Color cameras see a flat picture. Deptrum's cameras see distance. That difference is the entire business.
Deptrum, known in China as 光鉴科技, is headquartered in Shanghai with a second base in Shenzhen. It is not a household name, which is rather the point. Its work hides inside other companies' products: a robot's eyes, a cockpit's attention monitor, a turnstile's biometric lock. You are meant to never notice it. You are only meant to notice when it fails, and the bet is that it does not.
Color cameras tell you what something looks like. Depth cameras tell you where it is. Only one of those keeps a robot from walking into a wall.
By the late 2010s, the software for machine intelligence was racing ahead. The senses were lagging. A robot could run a neural network that recognized a thousand objects and still trip over a step it could not measure. Depth - the plain fact of how far away things are - was the missing input.
The hardware that supplied it was either expensive, bulky, or controlled by a handful of foreign chipmakers. Structured-light modules leaned on VCSEL laser arrays. LiDAR was accurate and costly. For a Chinese device maker who wanted reliable 3D sensing at scale, the options were thin and the bill was high.
Deptrum's founders looked at that gap and saw a supply problem dressed up as a physics problem. The depth was achievable. The trouble was that no one had made it small, cheap and home-grown all at once.
So the question they set out to answer was narrow and stubborn: could you rebuild the optics from the chip up - shaping light itself with nanophotonics - and skip the parts everyone assumed were mandatory? If yes, depth sensing could go from a specialty component to something you put in everything.
The depth was never the hard part. Making it small enough, cheap enough and yours enough - that was the hard part.
Deptrum was founded in 2018 by three people who had spent their careers bending photons for a living. Their bet was specific: that nanophotonics - engineering light at a scale smaller than its own wavelength - could replace bulkier depth optics and produce the world's first structured-light system that did not need a VCSEL array. It was a contrarian wager, and it has held up enough to draw both investors and lawsuits.
Ph.D. in Electrical Engineering from UC Berkeley, B.S. from Tsinghua. Before Deptrum, he led depth-sensing hardware design at Apple. Forbes China 30 Under 30, 2018. His Berkeley work on artificial chameleon skin was named Optics of the Year in 2015.
Ph.D. in Electrical Engineering from Stanford University. Runs the technology stack that turns nanophotonic optics into shippable depth modules.
Ph.D. in Electrical Engineering from UC Berkeley. The research conscience of the company, anchoring its optics-first approach.
A founder who led depth hardware at Apple and a chameleon-skin paper named Optics of the Year. The pedigree is the pitch deck.
Li Zhu's PhD work on artificial chameleon skin earns one of optics' headline honors - the technical seed of what comes next.
Three Silicon Valley scientists, including a former Apple depth-hardware lead, start the company in China. Zhu makes Forbes 30 Under 30.
Deptrum ships what it calls the world's first non-VCSEL structured-light depth solution, built on its Wavefront Processor.
Stellar (ToF), Aurora (structured light) and Nebula (sToF fusion) lines roll out for robotics, retail and industry.
A round backed by investors including CICC Capital fuels expansion. Total funding climbs toward $68M across six rounds.
HandPass palm-recognition cameras push into biometric payments; automotive-grade ToF moves into smart cockpits.
Deptrum's catalog reads like an astronomy syllabus, which is either a branding decision or a tell about how the founders think. Underneath the sky-themed names sits one common idea: a nanophotonic Wavefront Processor that shapes light, paired with AI algorithms that turn raw signal into usable depth. The company sells the whole stack - optics, sensor, chip and software - rather than a single piece.
High-precision structured-light cameras (900/910/930) for robot obstacle avoidance and volume measurement, working from dim rooms to 80,000 lux.
ToF depth cameras (400/420/800) with on-board ARM processing for robots, smart homes and security systems.
Structured-ToF hybrids (400/240) that pair near-distance ultra-high precision with far-distance depth in one module.
Dual-mode palm-recognition camera reading palm-print and vein, with depth-based anti-spoofing for contactless payment.
A structured-light face-unlock module that hides beneath a smartphone display - secure recognition you cannot see.
The nanophotonic chip underneath it all, enabling structured light without a VCSEL laser array.
Every product above answers the same question - "how far away is that?" - in a slightly different accent.
They named the cameras after stars and nebulae, then pointed them at the most earthbound jobs imaginable: a robot's next step, a palm at a checkout.
Talk is cheap in deep tech; the receipts are the patents, the rounds, and who bothers to sue you. Deptrum has filed more than 1,000 patent applications and collected 50-plus industry honors. It has raised roughly $68 million across six rounds, capped by a $28 million Series B in 2023 backed by investors including CICC Capital. And in a backhanded compliment, rival Orbbec sued Deptrum for about $7 million in a patent dispute - the kind of attention reserved for companies a competitor has started to take seriously.
A chart is just a sentence wearing a suit. Translation: well-funded, heavily patented, and worth suing.
Where do the cameras end up? In financial payment terminals, intelligent robots, automotive smart cockpits, smartphones and new-retail systems. Deptrum sells to the makers of those devices, not to you - a classic B2B arrangement where success means being invisible to the end user and indispensable to the integrator.
You can measure a deep-tech company by who sues it. Orbbec's $7M patent claim was, in its way, a five-star review.
Deptrum's stated mission - "Creating New Dimensions," building visual infrastructure for the IoT era - is the kind of line that sounds like a poster until you notice it is also a roadmap. The argument runs like this: the number of machines that need to perceive physical space is growing fast. Robots, vehicles, payment systems, appliances. Each one needs depth. Today most of them do not have it, or pay too much for it.
If depth sensing follows the path of the regular camera - from rare and expensive to cheap and ubiquitous - then whoever makes the small, affordable, home-grown 3D module is selling picks during a gold rush. That is the position Deptrum is trying to hold: not the flashy robot, not the famous car, but the sensor every one of them quietly depends on.
Creating New Dimensions. Building visual infrastructure for the IoT era.
Before founding Deptrum, CEO Li Zhu led depth-sensing hardware design at Apple - the same family of tech behind Face ID.
His UC Berkeley PhD invented an "artificial chameleon skin," named Optics of the Year in 2015.
Deptrum claims the world's first structured-light depth camera that works without a VCSEL laser array.
The product lines are all named after the sky: Stellar, Aurora and Nebula. The biometrics line, more bluntly, is called HandPass.
Rival Orbbec sued Deptrum for around $7M - a sign the startup became a genuine threat fast.
Return to where we started. The warehouse robot rounds the corner and does not clip the shelf. The car notices the driver's eyes drifting shut and chimes. The payment pad reads a living palm and waves a photograph away.
None of those moments look like technology. They look like nothing happening, which is the highest compliment a sensor can earn. The robot just works. The car just watches. The checkout just clears. The third dimension, once the missing input that left smart machines half-blind, has quietly become assumed.
That is the change Deptrum is wagering its existence on: a world where every machine that moves through space can measure it, cheaply, without anyone thinking twice. The company that wins that bet does not become famous. It becomes furniture - the thing everyone uses and no one mentions. Deptrum seems entirely content with that. The black bar does the seeing. You get to forget it is there.