All posts

Engineering

9 min

Inside the S1 perception stack

Twelve cameras, one lidar and a model that turns raw pixels into a live map of everything that matters on the floor.

Twelve eyes

S1 carries twelve cameras and a solid-state lidar. Together they cover every direction around the robot, with overlapping views that keep working when one sensor is blocked.

From pixels to a map

A perception model running on board turns those streams into a live 3D map thirty times a second, labelling people, pallets, shelves and anything S1 might need to pick up.

Built for change

Warehouses change hourly. Instead of relying on a fixed map, S1 rebuilds its understanding continuously, so moving a shelf never breaks a workflow.

Create a free website with Framer, the website builder loved by startups, designers and agencies.