monochrome

The problem

Most robots are nearly blind in the dark. Warehouses run overnight shifts, mines operate deep underground and agricultural machines work before dawn, yet standard cameras degrade exactly when automation would be most valuable. Teams compensate with expensive lidar, bright floodlights or human supervisors, and each workaround adds cost, energy use or risk. Low-light perception had quietly become one of the biggest barriers to round-the-clock automation.

Why now

Low-light perception was a research curiosity for years because the models were too heavy to run on a robot. That changed as specialised edge processors became cheap and efficient enough to run sophisticated neural networks in real time on battery power. Meanwhile, labour shortages pushed warehouses and mines to extend operating hours, and energy costs made lighting entire facilities overnight increasingly expensive. Robot makers needed perception that worked in any conditions without adding bulky sensors. Monochrome arrived just as the hardware made its approach practical and the market started actively asking for it.

The founders

Ava Zhang and Theo Lindqvist came out of a computational imaging lab where they had spent years teaching cameras to see what the human eye cannot. Ava focused on the neural models that reconstruct images from noisy sensor data; Theo designed the hardware pipeline that lets those models run on inexpensive chips. Together they had a rare combination: research-grade results and the engineering discipline to ship them in a product that costs less than the robot’s wheels.

Why we invested

Aria led our Series A after watching Monochrome’s system outperform a lidar rig inside a blacked-out loading dock. The demo was simple and convincing: their robot navigated pallets and people with the lights off, while the comparison system hesitated and stopped. Beyond the technology, Aria was impressed by how the founders had already signed design partnerships with hardware makers instead of trying to build their own robots. That focus on being the vision layer rather than the whole stack gave them a clear path to scale.

How we helped

After the round, we helped Monochrome formalise partnerships with four robotics manufacturers, using our network to arrange introductions with engineering leaders rather than procurement teams. Aria worked closely with Ava on hiring, helping them recruit a head of developer relations who made their SDK far easier for partners to integrate. We also introduced the team to a contract manufacturer that cut their sensor module cost by almost forty percent.

Where they are now

Monochrome’s vision stack now ships inside robots from four hardware partners and runs in warehouses, mines and greenhouses across three continents. Customers report running full lights-off shifts for the first time, saving energy and extending operating hours. The team is now expanding into outdoor agriculture, where dawn and dusk work have always been difficult for machines. Ava still sends Aria a short clip every time a new customer runs its first fully dark shift.