NSF SBIR Phase I · Active Development

AgriMesh

Offline-first precision agriculture for urban micro-farms.

AgriMesh is a network of autonomous robotic micro-farm nodes — each capable of sensing, deciding, and acting without Internet. Built for food deserts, schools, and community farms where broadband is unreliable and staffing is limited.

Precision agriculture assumes what urban communities don't have.

"When connectivity fails, cloud-dependent monitoring and analytics can degrade exactly when reliability matters most."

Urban food deserts and under-resourced community agriculture programs face a practical constraint that most precision-ag systems assume away: intermittent connectivity, uneven staffing, and limited capacity to monitor dispersed sites.

AgriMesh is designed for the opposite assumption: intermittent backhaul is normal, and micro-farm nodes must continue operating safely and usefully without it.

1 in 5

Miami-Dade residents face food insecurity

40%

of urban land sits underused or vacant

0%

broadband required for AgriMesh node operation

20%+

target water savings vs. rule-based baseline

Three proven components. One new systems-level capability.

AgriMesh integrates robotic actuation, edge compute, and self-healing mesh networking into a single architecture purpose-built for urban micro-farms.

Robotic Actuation Layer

FarmBot-class robotic gardening platform executes repeatable micro-farm operations — watering, seeding, cultivation — with a consistent physical interface for sensors and implements.

Edge Compute Layer

Each node runs a local data pipeline: sensor ingestion, validation, storage, scheduling, and safe actuation. Localized inference runs on-device via embedded ML runtimes designed for <256KB RAM.

Mesh Networking Layer

Nodes form a secure, self-healing mesh using established protocols (Babel routing, batman-adv) — no central cloud required. An optional gateway provides backhaul when available.

Modular Resource Resilience

Optional solar backup and rainwater capture reduce OPEX and enable deployments on lots without convenient utilities — tested in Phase I, designed for Phase II scale.

Sensors

Soil · Weather · Vision

Edge Compute

Inference · Storage · Control

FarmBot

Actuation · Irrigation · Tools

Mesh Network

Telemetry · Commands · Sync

Three measurable objectives. Quantitative pass/fail metrics.

Offline-First Closed-Loop Control

A

Each AgriMesh Node autonomously manages irrigation and safety checks when connectivity is lost — no cloud dependency required. The sensor-to-actuation pipeline runs entirely on-device.

≥95% successful actuation events over a 30-day continuous trial

≥98% telemetry data captured and stored locally

Verified safe failover (e-stop, irrigation limits) under outage scenarios

Mesh Networking for Micro-Farm Fleets

B

A self-healing neighborhood mesh connects multiple farm nodes, sharing telemetry and commands across dispersed sites — even when Internet backhaul is intermittent or absent.

p95 command latency ≤250 ms across 2 hops

≥97% node reachability uptime over 30 days

Network reconvergence ≤60 seconds after node failure

Localized Agronomic Inference

C

Edge AI modules run directly on each node — optimizing irrigation schedules and detecting sensor or valve faults without sending data to the cloud.

≥20% water savings vs. rule-based baseline

≤5% yield difference vs. baseline

Anomaly detection: ≥90% precision / ≥80% recall on injected faults

12-month roadmap from prototype to field validation.

M1–M2

Preparations

System requirements & test protocols

Threat model & security design

M3–M7

Development

Node integration (sensors + compute)

Control logic + safety features

Mesh stack selection & bench tests

Edge inference MVP + baseline

M8–M10

Field Trial

Miami-Dade pilot deploy & run

Performance evaluation & iteration

M10–M12

Commercial Prep

Buyer interviews & pricing studies

Unit-economics model & Phase II plan

M2

Requirements doc & test plan

M5

Prototype node v0.5

M6

Mesh benchmark report

M9

Edge-inference results

M11

30-day field data (Miami-Dade)

M12

Feasibility report + financial model

Organizations that already run urban farms — but struggle with labor and reliability.

AgriMesh's value proposition is clear: reduced labor per bed, improved yield stability, reduced water use, and fleet-level visibility across dispersed micro-sites. Competing solutions assume broadband and centralized compute. AgriMesh does not.

Municipal / Public Health Programs

City-run urban farm programs and community garden initiatives seeking reduced labor and guaranteed uptime.

School Districts / CTE Programs

STEM and Career & Technical Education programs that need reliable, low-maintenance farm infrastructure.

Nonprofits / CDCs

Community development corporations and food-access nonprofits managing dispersed micro-farm sites.

Housing Authorities

Public housing garden initiatives requiring automated, low-staffing food production at scale.

Hardware + subscription + services.

Hardware / SW Licensing

Node kits (robot + sensors + mesh hardware) and installation. Estimated ~$8K–$10K in materials per node.

Subscription (Ops)

Cloud dashboard, analytics, updates, and support per node. Target pricing: $300–$600/node/month, validated through Phase I customer interviews.

Services

Training, curriculum materials for schools and community programs, and maintenance contracts.

Illustrative 3-year payback model: ~$10K node cost + installation, $400/month subscription → break-even at ~Month 25. Phase I refines these figures with market feedback from ~20 buyer interviews.

McCartney's Concentric Company, Miami, FL.

A Miami-based agri-tech integration firm (est. 2014) specializing in systems integration across embedded software, networking, and micro-farm operations. Phase I leverages a multi-disciplinary team including embedded developers, an agronomy consultant, and a network security advisor.

AgriMesh is incubated within the FarmBlock ecosystem, with an existing Miami-Dade pilot partnership in Liberty City and Overtown — where micro-farm nodes can be deployed and evaluated under realistic urban constraints.

2014, Miami, FL

Liberty City / Overtown

3–5 nodes, Phase I

NSF SBIR Phase I

Security & Privacy

No private human data is collected. Nodes log only environmental and system telemetry. All mesh communications are encrypted; each node uses unique credentials. Encrypted at-rest storage and TLS links mitigate common risks.

Sustainability

Solar panels with battery backup and rainwater catchment are tested on 1–2 Phase I nodes, designed per EPA guidelines for non-potable irrigation use. These resilience modules are optional in Phase I, foundational in Phase II.

Phase II Roadmap

Phase II expands to container/indoor farms, automated greenhouses, and optional seed-to-fork traceability via a DAG-based tamper-evident log — contingent on validated customer requirements for auditability.

Interested in partnering, piloting, or funding AgriMesh?

We are actively seeking pilot site partners, letters of commitment from school districts and municipal programs, and research collaborators. Phase I buyer interviews are open.