Vegard

Autonomous Agricultural Intelligence Platform — Drone Fleet Coordination & Soil Intelligence Fabric

Vegard is the autonomous systems coordination layer for the Enotrium agricultural stack. Analogous to Anduril Lattice for defense, Vegard coordinates fleets of UAVs carrying hyperspectral sensors, fuses their data into a shared operational picture, and routes soil intelligence to the AIP supply chain platform.

Quick Start

# Install
pip install -e .
pip install -e ../arthedain  # Optional: edge learning dependency

# Start Vegard API
python -m vegard.command.api

# Simulate a fleet of 4 drones
python sandbox/simulate_fleet.py --drones 4 --duration 60

# Test AIP integration
python -m sandbox.mock_aip_server --port 3000 &
python sandbox/test_aip_bridge.py --aip-url http://localhost:3000

Architecture

Vegard is structured in three layers:

Layer 3 — Node (Edge Intelligence)

• NodeAgent: Drone ↔ fabric bridge, executes scan tasks
• SpectralBridge: Arthedain SNN → Hyperspectral-Restruct CNN adapter
• Integrations: Arthedain client, HSI API client with graceful mock fallbacks

Layer 2 — Fabric (Coordination)

• Mesh: Gossip protocol state mesh
• TaskAllocator: Auction-based tasking with preemption
• DriftMonitor: Cross-node drift correlation (key differentiator)
• Attestation: PGP-signed soil outputs for AIP
• Transport: gRPC (drone↔node) + MQTT (cloud↔AIP)

Layer 1 — Command (Operational Picture)

• FusedFieldPicture: Materialized view, GeoJSON export for Mapbox
• MissionPlanner: Goal-to-task conversion, priority scoring
• AIPBridge: Clean POST to /api/vegard/ingest, zero circular coupling
• API: REST + WebSocket operator interface

Key Features

• Spatial Drift Correlation: Detects contamination events by correlating Arthedain E(t) signals across drones
• Auction-Based Tasking: Distributed task allocation without central coordination
• Attestation: PGP-signed predictions for supply chain provenance
• Multi-Protocol Transport: gRPC for performance, MQTT for compatibility
• Sandbox Mode: Full simulation without live hardware

Repository Structure

EnotriumSyndicate/Vegard/
├── vegard/
│   ├── node/              # Edge intelligence bridge
│   ├── fabric/            # Coordination services
│   ├── command/           # Operational picture + AIP integration
│   └── proto/             # Protocol Buffer definitions
├── configs/               # YAML configurations + test polygons
├── sandbox/               # Simulators + mock services
├── hardware/              # FPGA interface spec
├── tests/                 # Test suite
├── pyproject.toml         # Package configuration
└── CLAUDE.md              # Full architecture documentation

Dependencies

• Python 3.11+
• gRPC, Protobuf, MQTT
• FastAPI, WebSockets
• NumPy, Shapely
• gnupg (for attestation)
• arthedain (optional, local editable install)

Documentation

• Full architecture: CLAUDE.md
• FPGA interface: hardware/fpga_interface.md
• API reference: GET /docs when running API server

The Enotrium Stack

Repo	Role
arthedain	Edge SNN learning algorithm
Hyperspectral-Restruct	3D CNN soil prediction model
Vegard	Fleet coordination
AIP	Supply chain command center
Icarus	Drone hardware platform (planned)

License

MIT — Enotrium Syndicate Vegard