Skip to content

Example Gallery

Use the examples in this order. The core repo teaches the runtime itself; GoldenRetriever carries larger perception, memory, language, robotics, and integration examples.

01 First visual demo

Webcam frames flow through a color detector and display path.

pixi run demo-webcam-detection 02 Core mechanics

Typed payloads, Flow classes, clocks, sync policies, and pipeline wiring.

pixi run demo-basic-flow 03 Golden examples

Perception, memory, language, notebooks, and robotics integration lanes.

pixi run -e golden-local ...

Core Runtime Examples

These examples are intentionally small. They are meant to make the runtime model obvious before you bring in heavier robot stacks.

Goal Command What it teaches
See a graph run on real input pixi run demo-webcam-detection Camera input, detector Flow, visualization fallback.
Understand the minimum API pixi run demo-basic-flow @io, Flow, Rate, Trigger, Pipeline.connect.
Inspect backend execution pixi run demo-rt-execution Runtime validation and backend execution.
Debug in one Python process pixi run demo-stepper Pipeline.step(...) before multiprocessing/dora.
Record and replay perception pixi run demo-webcam-record Deterministic replay artifacts for debugging.
Join multi-rate streams pixi run demo-data-multistream-join Synchronization and event joins.

Core vs GoldenRetriever

  • Keep examples in this repo when they teach the runtime contract: typed flows, clocks, sync, IR, execution, stepping, replay, or release checks.
  • Put examples in GoldenRetriever when they teach a robotics application lane: perception models, memory, language grounding, notebooks, simulation, real robot adapters, or larger system demos.
  • Link across repos instead of copying code when the same concept would otherwise appear twice.

Current boundary

No example move is needed for the current docs polish pass. The core repo already has the right tutorial primitives; GoldenRetriever remains the right home for larger robot-facing examples.

Next Paths