Moseley Climacteric™

01 · The system

Moseley Climacteric™ is an integrated biological intelligence platform for fresh produce environments, combining distributed sensing, predictive machine learning, and active environmental remediation within a single closed-loop architecture.

The platform is designed to detect, interpret, and respond to produce respiration and ethylene-response behaviour in real time — from pallet-level biological sensing through to cloud-based trajectory prediction and in-environment remediation.

Three layers. Eight components. One continuous intelligence loop — from the first detectable biological change within a shipment to predictive spoilage analysis and active environmental intervention before product quality is lost.

New to ethylene? Start here · 4:41

What is ethylene, and how do you sense it?

A plain-language introduction — no chemistry assumed. Learn what ethylene is, the role it plays in ripening fresh produce, and how researchers at Henan Agricultural University built a biomimetic copy of the plant's own ethylene receptor — then turned that molecular capture into a measurable signal that powers the Moseley Climacteric platform.

Watch · 3:48

The Moseley Climacteric Platform, at a glance.

A narrated overview of the platform architecture — from biomimetic ethylene sensing and distributed E-Nose telemetry, through predictive ML trajectory modelling, to coordinated remediation across reefers, cool rooms, distribution centres, and retail environments.

▶ Platform overview · 3:51

Reading the chart · Ethylene trajectory over 7 days

A single carton, watched continuously — from baseline to remediation.

This chart shows what Moseley sees inside one climacteric carton over a seven-day cold-chain window. The y-axis is ethylene concentration in parts-per-billion (ppb); the x-axis is time in days. The faint trace is the raw 1 Hz sensor reading from the metal-cystine sensing element at 4 °C. The bold cyan line is the predicted ripening trajectory generated by the on-device E-Sentinel model — a forward-looking forecast, not just a measurement.

Three moments matter. The detection threshold (~20 ppb, dashed line) is when the platform first calls a biological event. The predict-ahead window (highlighted) is where the model says “this carton will tip into accelerated ripening within the next few hours.” The E-MCP trigger (arrow) is the moment the active remediation node is dosed — ahead of the curve, not after the damage. The downslope that follows is the ethylene decay once the MOF canister is pulling gas out of the airspace.

D0 – D1

Baseline

Sensor sits below the 20 ppb detection threshold. Produce is metabolically quiet — the platform is listening, but there is nothing to act on yet.

D1 – D3

Climacteric onset

Ethylene begins to accumulate as autocatalytic ripening kicks in. The raw signal climbs above threshold and the trajectory model locks on to a rising curve.

D3 – D4

Predict-ahead window

E-Sentinel forecasts the carton will exceed the spoilage envelope inside the next forecast horizon. The platform calls for active intervention before peak ethylene, not after.

D4 – D7

E-MCP triggered → decay

The MOF canister is dosed into the airspace. Ethylene falls from ~320 ppb back toward ~60 ppb over three days — the curve the buyer never sees, because the fruit holds its quality.

Reading the diagram · System architecture

Three lanes, one closed loop — from the pallet to the cloud and back.

Moseley Climacteric is built as three horizontal lanes that flow left-to-right and converge on a single cloud hub. Detect at the top is the sensing layer that lives with the produce. Predict in the middle is the machine-learning layer that turns raw signal into a ripening forecast. Remediate at the bottom is the actuation layer that physically removes ethylene from the airspace. All three lanes feed the E-Platform cloud, which fuses the data, drives operator dashboards and APIs, and closes the loop back to remediation with a single autonomous control signal.

Read the diagram as a journey: a pallet of produce passes a sticker-class E-Nose Label as it moves through the cold chain; the host E-Aegis platform aggregates that telemetry; the E-Sentinel TinyML model running on-device computes a calibrated risk score; the E-Platform cloud orchestrates a fleet of these nodes and — when risk crosses threshold — triggers an E-Remediation canister to act in the same airspace as the fruit. No human in the loop, no shipment-level guesswork.

Moseley Climacteric Platform architecture — DETECT, PREDICT, REMEDIATE closed loop through the E-Platform cloud — Detect · Predict · Remediate — a closed loop from molecular sensing, through edge ML forecasting, to autonomous filtration acting in the same airspace as the fruit.

01 · Detect

The sensing layer

Biomimetic ethylene sensing at the pallet. E-Nose Pallet Label is a sticker-class sensor; E-MCP is the release node; E-Aegis is the host platform that runs them. This is what lives with the fruit.

02 · Predict

The intelligence layer

E-Sentinel is the on-device TinyML model. It turns raw sensor readings into a ripening trajectory and a calibrated risk score — so the system acts on forecast, not lag.

03 · Remediate

The intervention layer

E-Remediation is the MOF-based ethylene-removal canister. When the risk score crosses threshold, the canister is dosed into the same airspace as the produce — holding ethylene below the ripening envelope.

Hub · E-Platform

The closed loop

All three lanes converge on the cloud. The platform fuses fleet telemetry, drives operator dashboards, alerts and APIs, and — critically — sends the autonomous control signal back to remediation. That arrow back is what makes it a loop, not a dashboard.

01 · Detect

Distributed change-detection sensing

The sensing layer does not chase an absolute ethylene concentration — it watches for the rise. A biomimetic chemistry stack built on Cu / Fe / Zn / Co / Ni / Mn metal-cystine coordination motifs responds to small, early increases in ethylene against a stable baseline of other ripening volatiles. Water-vapour drift is compensated; an ethylene-selective filter layer narrows the response. The output is a trend signal — exactly the input the trajectory model needs.

02 · Predict

Trajectory inference, not just measurement

An ML model runs first on the sensor board and again in the cloud, consuming the change signal from the array and fusing it with the other volatile channels into a ripening trajectory and a remaining shelf-life estimate. The biomimetic chemistry layer cross-validates ethylene specificity through multiple metal-cystine coordination motifs in parallel.

03 · Remediate

Active scrubbing in the same airspace

When the predicted ethylene trajectory crosses a threshold, a multi-bed adsorbent scrubber is gated on by the trajectory model. Remediation happens in the same airspace as the fruit, with no operator intervention — a closed loop from sensing to actuation.

00 · Platform

The cloud that ties it together

The platform ingests every sensor stream across the fleet, runs cloud-side fusion against a reference instrument set, drives operator dashboards and alerts, and pushes model updates and gating thresholds back to the edge. Without the platform there is no fleet — only isolated nodes.

“Climacteric” is the botany term for the post-harvest ripening burst triggered by ethylene. The product name takes that biological reality as its starting point.

02 · The hardware

One device. Three modifiable layers.

The Moseley E-Aegis is a single self-powered device that draws an air sample, conditions it, senses ethylene and accompanying contaminants down to the parts-per-billion range, decides what to do about it, and remediates — all on board, all under machine-learning gating. Internally it is built as three independent layers — E-Conditioning, E-Array and E-Remediation — each of which can be modified or substituted to suit the environment (cool room, refrigerated trailer, sea container, distribution centre).

▶ E-Aegis explainer · 0:52

E-Aegis · sample → condition → sense → decide → remediate

E-Aegis

Predictive Environmental Sensing and Autonomous Remediation System.

A single ambient-energy-augmented unit that lives inside the cargo space for the whole journey. An inlet draws a sample. The conditioning stack strips interferents. The sensing array reports ethylene and related VOC trajectories. The on-device machine-learning model decides when to gate the remediation stack. The unit transmits a verified trajectory upstream to the Moseley cloud and operates for not less than 42 days without external charging. The three layers below are independent — each can be modified, swapped, or tuned to suit the deployment.

TargetsC₂H₄, H₂O, NH₃, VOCs

Sensitivity0.01–10 ppm, LOD ~1 ppb

Decision layerOn-device ML · ML-gated remediation

PowerAmbient-energy harvester + battery · ≥ 42 day floor

ConnectivityBLE 5.3 · LoRa optional · gateway upstream

Operating range−20 °C to +40 °C

ComplianceCE · FCC · RoHS · IATA-safe

E-Aegis architecture · the three layers

Inside the E-Aegis: condition, sense, remediate.

The E-Aegis is built as three independent layers that work as one device — each tunable for the deployment without redesigning the others.

Layer 1 · E-Conditioning

Multi-stage sample preparation — filter, desiccate, O₂ scrub, S guard — tuned per environment before any sensing happens.

Layer 2 · E-Array

Metal-cystine sensing array with ML fusion — turns a multi-channel response into a single confidence-scored ethylene trajectory.

Layer 3 · E-Remediation

Sentinel-gated multi-bed receptor-in-host capture — beds engage only when the model warrants it, conserving sorbent life.

See the full E-Aegis architecture — specs, layer images and audio explainers →

Moseley E-MCP — distributed ethylene-modulator release node, three-quarter view with top louvre vent, carbon-fibre band, glowing teal accent ring and photovoltaic sidewall

E-MCP · distributed controlled-release node · deploys on top of pallets inside the cargo space

E-MCP · methylcyclopropene release device

The actuator arm of the architecture — distributed, ML‑gated ethylene-modulator release.

The E-MCP is a self-powered hockey-puck-sized release node that sits on top of pallets inside refrigerated containers, cold rooms and ripening rooms. A sealed cartridge holds an ethylene-action payload bound to a metal-organic-framework matrix; a heater beneath the cartridge releases payload on demand, a rotating-disc shutter and condensation-protected vent dispense vapour into the headspace, and a flat battery pack augmented by an organic-photovoltaic sidewall delivers a 42‑day operational floor. A sealed, periodically-gated sensing cavity — carrying a MOF + biomimetic ethylene receptor and a gravimetric sensing element such as a quartz crystal microbalance — lets the puck read its own headspace on a duty cycle and self-trigger release through on-device TinyML inference. Nodes operate autonomously or under E-Aegis orchestration, with release events gated by predicted climacteric progression — never raw ppb.

▶ E-MCP explainer

▶ What is MCP?

FunctionDistributed on-demand ethylene-modulator release · ML-gated · self-sensing

Form factorCylindrical puck · ~80 mm × 40–50 mm

Release stackMOF-bound payload cartridge · heater · rotating-disc shutter · louvre vent

Self-sensingSealed gated cavity · MOF + biomimetic ethylene receptor · gravimetric sensing element (QCM) · on-device TinyML inference

Moisture isolationSealed shutter chamber · hydrophobic membrane · condensation trap

PowerFlat battery pack + organic-PV sidewall · ≥ 42 day floor

Operating modeAutonomous · or orchestrated under E-Aegis

ConnectivityBLE 5.3 · NFC cartridge identity

Integration with architectureDistributed actuator · gated by E-Sentinel ML · coordinated by E-Aegis

Full E-MCP product page →

+ E-MCP · In the field

A reefer container, instrumented end-to-end.

Several E-MCP nodes sit on top of the pallet stacks. An E-Aegis sentinel at the front bulkhead coordinates release events across the load. Reefer return-air carries each puck’s controlled vapour egress through the headspace so dose is delivered to identified ripening hotspots rather than the load as a whole.

Cutaway scene inside a refrigerated shipping container with five E-MCP pucks on top of palletised cardboard cargo and teal vapour streams rising into the headspace, with dotted BLE coordination rays from an E-Aegis sentinel on the front bulkhead to each puck

Distributed E-MCP nodes in a reefer container · orchestrated under E-Aegis · vapour egress entrained into the return-air path

E-Nose Ethylene

A biological sensing label for every pallet.

An ultra-thin intelligent sensing patch designed to continuously monitor ethylene-response behaviour and produce headspace conditions throughout the cold chain. Attach directly to cartons or pallets and the platform captures biological telemetry across the entire journey using infrastructure-free opportunistic communications.

▶ E-Nose Label explainer · 0:53

Form factor68 × 68 × 1.4 mm patch

Weight9 g

TargetEthylene (C₂H₄)

Sensitivity25 ppb

Sample rate0.1 Hz · averaging mode

BatteryPrinted Zn-air cell · 30-day life

ConnectivityBLE 5.3 advertising

Operating range0 °C to +35 °C, 10–95% RH

ActivationQR scan · tap-to-pair NFC

ComplianceCE · FCC · food-contact safe

+ Predict · the on-device intelligence

E-Sentinel.

ME-S200 · Edge prediction engine

A TinyML inference engine running on the same PCBA as the E-Nose Label and inside the E-Canister. It classifies climacteric state — Baseline → Early → Confirmed → Critical — from the rate of change of ethylene emission measured across a metal-cystine sensor array, fused with humidity, CO₂ and temperature signals. No cloud round-trip required. Models update over the air via the fleet-learning loop.

Architecture

Teacher–student TinyML

Inference

Microcontroller-class, on-device

Calibration truth

Metal-cystine sensor array (AU 2026904494)

Output

4-state classification + trajectory

Latency

Real-time, no cloud round-trip

Update path

OTA via fleet-learning loop

Runs on

E-Nose Label · E-Canister

Stack

StiknTrak comms backbone

**Predictive Biological Intelligence for Fresh Produce.**