The mesh
ResQ runs as a decentralized mesh, not a hub-and-spoke cloud. Drones, ground units, and operator stations form a peer network that keeps operating when upstream infrastructure is down or unreachable.- Nodes communicate over local-first transports (mesh radio, LTE, Wi-Fi) and reconcile with the cloud opportunistically.
- The Coordination API is designed to keep accepting telemetry and serving live state when its upstream dependencies are degraded.
- There is no single point of failure. If a coordinator drops, peer nodes continue to share telemetry and queue work.
503 Service Unavailable for some
routes during partial outages — see Errors for retry guidance.
Evidence and the chain
Every consequential action in a mission produces evidence:- Drones capture sensor frames, video, and structured telemetry.
- Evidence files are pinned to IPFS and referenced by their CID (content-addressed identifier).
- CIDs are anchored on Solana, producing a tamper-evident chain.
- The Infrastructure API exposes both halves:
/evidencefor the IPFS payload,/blockchain/*for the on-chain anchors.
Human-in-the-loop missions
ResQ’s autonomous flows are HITL-gated. The platform implements EU AI Act Article 14 human oversight: an authorized operator must approve high-risk actions before the system commits. Mission approval is exposed via the Coordination API:GET /admin/missions/pending— actions awaiting approvalPOST /admin/missions/approve— green-light a pending missionPOST /admin/missions/reject— block and record the reason
missions.approve operator scope. Calls without it
return 403 — see Errors.
Airspace and permits
For deliveries and autonomous flight, ResQ uses an on-chain airspace registry on Solana. The Infrastructure API’s/solana endpoints record
permits, delivery events, and registry queries. The dispatcher refuses
flight plans that fall outside permitted airspace; this gate sits in
front of mission approval, not after it.
Telemetry and live events
Two flows carry real-time data:- Ingestion: drone fleets push telemetry batches to
POST /fleet/telemetryon the Coordination API. Batches are buffered on the edge and retried — telemetry never silently drops. - Subscription: clients consume live state via Server-Sent Events at
/eventsand Prometheus scrapes at/metrics(Coordination API).
Operator identity and scopes
Operators authenticate with username + password atPOST /login and
receive a short-lived JWT (see Authentication). The
token carries the operator’s scopes — fine-grained permissions like
missions.approve, evidence.write, or airspace.admin.
A request that succeeds in authentication but lacks the required scope
returns 403. Surface this to the operator and do not retry; it requires
an administrator action.
Fault injection and simulation
The Coordination API exposesSimulation endpoints for fault injection,
and the SDKs ship simulation harnesses. The intent is to exercise the
mesh’s degraded-mode behavior in test before depending on it in
production. Use it during integration tests, not in live operations.
What’s where
Next
Quickstart
First authenticated call.
Authentication
JWT lifecycle and scopes.
API reference
Every endpoint.