Conceptual architecture
A closed loop between observation and action.
Every decision starts from a real measurement and passes through a safety controller before acting on the environment.
Sensors
Temperature, humidity, substrate moisture, light, CO₂.
Data collection
Continuous, timestamped acquisition of measurements.
Environment analysis
Putting measurements in context against the crop's needs.
AI agent
Proposing agronomic actions based on the analysis.
Safety controller
Deterministic validation of actions before execution.
Actuators
Irrigation, ventilation, lighting, heating.
Plant
The crop responds to the actions applied.
New observation
The cycle restarts from new measurements.
The loop restarts with every new observation: measure, decide, act.
Safety
An AI that proposes, a deterministic controller that validates.
Verdant's AI agent analyses the environment and proposes actions. Those proposals pass through a local safety controller, built on deterministic logic, before any actuator is driven. This separation aims to ensure that no uncertain decision is applied directly to the physical world.
Targeted capabilities
What Verdant's architecture is designed to cover.
As the project is in development, some capabilities are operational and others are still being worked on.
- TemperatureOperational
- HumidityOperational
- Substrate moistureOperational
- LightOperational
- CO₂In development
- Measurement historyOperational
- Traceable AI decisionsOperational
- IrrigationOperational
- VentilationIn development
- LightingOperational
- HeatingIn development
- VisionIn development
- Local safetyOperational
- Web-based supervisionOperational