Sorter V2 Documentation

⚠ Under heavy construction

As of July 11, 2026, Sorter V2 is not yet in a position to be built.

The documentation that exists is incomplete. Any given page may be accurate, inaccurate, present only as an example, or badly out of date. We do not yet recommend that anyone attempt to build Sorter.

For the most live updates on our progress, join our Discord ↗.

Local Machine Software

SorterOS

SorterOS is the Sorter V2 local software, the Python backend that drives the hardware and the SvelteKit UI that operates it. Install it, operate it, and read how it works under the hood.

Install

  • Installation — the three routes: the SorterOS image for the Orange Pi 5, the one-command installer for generic Linux, or the sequence by hand.

Set up a new machine, in this order

  1. First setup in the UI. The setup wizard: name the machine, find the boards, check motion and endstops, assign the servos and the cameras, link Hive.
  2. Camera calibration. Focus each camera against a printed Siemens Star.
  3. Chute calibration. Home the chute, capture two bins to set the bin locations, and test every bin it can reach.
  4. Before your first sort run. The last five settings to check in the UI.
  5. Preparing LEGO for a sort run. What to take out of a tub of bulk LEGO before it goes in the bulk bucket, and what each thing does to the machine if it stays in.
  6. Your first sort run. The end-to-end happy path: pick a profile, feed the machine, check a bin, stop cleanly.

Operate

Under the hood

  • Dev flow — the dev backend and UI services, how to enable them, and the difference between a soft restart and a full restart.
  • Sorting profile reference — the on-disk shape of sorting_profile.json — rules, conditions, part_to_category, set inventories. Accurate for schema_version: 1.
  • machine.toml reference — every field in the machine-specific config file: servo, chute, carousel, stepper bindings and overrides, cameras, and GPIO LEDs. This lives in the docs site, not the SorterOS UI.

Coming soon

These flows do not yet have dedicated pages:

  • Operating the UI at scale — running long sessions, reviewing classification samples, tuning vision, checking runtime health.

Until those land, the authoritative sources are software/README.md and software/sorter/backend/coordinator.py.