The core nodes are the spine of the runtime: state estimation feeds the policy, the policy produces an action, the action publisher translates it into a MAVROS command, and takeoff_node handles the cold-start sequence from arm to first hover. If any one of these is broken, the drone is sitting on the ground.
Two pages here are dev-log-grade rather than reference-grade and worth reading in order: Takeoff node evolution v1–v5 walks through five iterations of getting the takeoff sequence right (and the four design rules we extracted from the failures), while Takeoff node v7 — event-driven FSM is the final architecture that survived --mavros integration. If you only have time for one, read v7 first; the evolution page is the why behind the what.
The other two pages — state_estimator_node and action_publisher_node — are currently stubs, because the corresponding components are still in design. They’ll get filled out when the implementations land. For now, the conceptual placeholders explain what each will do and which topics they’ll touch.
Contents
Auto-generated from child entries during build (update-indexes.mjs).