SYS / Ontology
The Physical World Ontology
Layer 1, and the reason everything above it works: the physical world modeled as intelligent objects rather than disconnected telemetry, every entity carrying relationships, ownership, location, capabilities, dependencies, state, constraints, and history.
SYS / What it models
Assets
UAVs, UGVs, robots, vehicles, sensors, batteries, chargers, with capabilities, dependencies, and state.
Places
Buildings, zones, airspace, roads, facilities, unified geometry and coordinate frames across every vendor.
Actors
People, operators, organizations, ownership, authority, and responsibility as first-class structure.
Operations
Missions, tasks, events, hazards, threats, inspections, failures, repairs, with schedules, histories, and constraints.
SYS / Predicting vs. reasoning
Context is not inferred from patterns. It is represented.
A standalone model sees
object_detected
class: vehicle · confidence: 0.87
lat: 34.9012 · lon: 33.6201
Kognilith sees
A damaged vehicle belonging to Facility A, inside restricted Zone B, operating below efficiency, with conditions raising failure probability. The assigned drone requires a battery whose charging station is unavailable.
Conclusion: the mission cannot continue as planned. Select an alternative asset, or delay.
This explicit structure is the difference between predicting and reasoning, and the reason the conclusions are reliable enough to act on.
SYS / How it works
A typed, versioned graph with a physics-aware core.
Three primitives anchor the graph: space, geometry, zones, and coordinate frames unified across vendors; time, histories, schedules, and predictions; and capability, what each asset can sense, carry, and do, under which conditions.
Asset adapters map each vendor's native data into these types at connection time. The hard translation work is done once per platform, in the adapter, against a stable semantic target, which is how integration collapses from months to hours.
At runtime the world model is the live state of the graph, updated by fused sensor input. Every reasoning agent operates by querying and writing it, plans, claims, and decisions are graph operations with full provenance. Every autonomous decision stores its reasoning chain as a path through the graph: the substrate of verifiable autonomy and the audit record regulators and commanders require.
SYS / Inside the ontology
-
01
Physical schema
Typed entities for assets, sensors, places, actors, and operations, with relationships, capabilities, constraints, and state as first-class structure.
-
02
Spatial core
Unified geometry, zones, and coordinate frames across every vendor's native system, one space all assets share.
-
03
Capability model
What each asset can sense, carry, and do, under which conditions, the basis of vendor-agnostic mission planning and task handoff.
-
04
Adapter mappings
Per-platform translations from native APIs into ontology types, integration as a solved, reusable artifact.
-
05
Rules & constraints layer
Airspace, regulations, safety envelopes, and organizational policies encoded as machine-checkable constraints enforced at planning time.
-
06
Mesh replication
Ontology subgraphs carried on-asset for disconnected reasoning, with principled reconciliation when connectivity returns.
-
07
Decision provenance
Every autonomous conclusion stored as its path through the graph, explainability and audit as native properties, not add-ons.