Domain / System Architecture

Layered System Architecture

Actuation and Mobility Layer

An autonomous wheeled platform navigates greenhouse rows while maintaining arm reach. A 4-DOF robotic arm performs dual intervention with interchangeable end-effectors: gripper for harvesting and nozzle for spraying.

Edge-Intelligence Layer

A Raspberry Pi 5 runs containerized YOLO inference for ripeness and foliar disease detection. The perception module provides coordinate outputs for targeted actuation and novelty alerts.

Resilient Networking Layer

ESP32 LoRa mesh nodes collect environmental data and deliver it to the edge gateway. The SDN control plane (ONOS + OVS) enforces adaptive QoS, while VPN-based failover ensures continuity during primary link instability.

Cross-Layer Data and Control Flow

The architecture follows a strict scan-detect-act-feedback sequence: sensing data and images enter the edge-intelligence layer, detections are converted into actuation coordinates, and SDN policies prioritize command packets so robotic actions complete even during network stress.

Design Goal

Ensure that high-precision mechanical actions are consistently supported by low-latency inference and reliable communication in rural greenhouse conditions.

Architecture Outcome

The integrated cyber-physical stack enables continuous autonomous operation beyond passive monitoring, including direct harvesting and disease treatment actions.

Back to Domain Hub