Quote of the day

What we know is a drop, what we don't know is an ocean.

- Isaac Newton

Contact

Edge AI and Causal Attribution for Cold-Chain Integrity in Agri-Food Export Corridors

Open Logistics & Supply Chain

Edge AI and Causal Attribution for Cold-Chain Integrity in Agri-Food Export Corridors

Edge AI and Causal Attribution for Cold-Chain Integrity in Agri-Food Export Corridors

Description

Detect cold-chain violations in real time using onboard sensors and identify the most likely causal mechanism behind thermal excursions along export corridors.

Details

Context and Problem Statement

Cold-chain failures cause major losses in perishable exports, but conventional loggers usually flag only threshold violations without identifying the root cause.

Research Question

Can a thermal anomaly be reliably attributed to an identifiable causal mechanism from multi-sensor edge data under energy and intermittent-connectivity constraints?

Proposed Approach

Instrument shipments with temperature, humidity, door-state, position, and refrigeration-power sensors. Detect anomalies using compact edge models, then perform causal attribution using a mechanism graph validated through controlled experiments. Estimate downstream shelf-life impact.

Expected Contribution

A shift from threshold alarms to experimentally validated causal attribution.

Expected Prototype

An autonomous tracking device and dashboard reporting likely cause, confidence, and estimated shelf-life impact for each incident.

Datasets

Public logistics sensor datasets, real export-route measurements, and cold-room experiments.

Challenges

Sensor autonomy, intermittent connectivity, experimental validation cost, and commercially sensitive traceability data.

Research Question

How can a cold-chain violation be attributed to an identifiable causal mechanism rather than merely flagged as a threshold exceedance?

Innovation

The project introduces experimentally validated causal attribution into a domain still dominated by threshold alarms.

Expected Deliverable

An instrumented tracking prototype, an attribution model validated by controlled experiments, and an evaluation on real logistics routes.

Technologies

Edge AI and TinyML IoT and Low-Power Protocols (LoRaWAN, NB-IoT) Time-Series Anomaly Detection Causal Attribution Product Shelf-Life Modeling

Required Skills

  • Embedded Systems and IoT
  • Signal Processing and Anomaly Detection
  • Applied Causal Inference
  • Experimental Design

Datasets

  • Public logistics sensor time series
  • Measurements from real export routes
  • Controlled cold-room experiments

Morocco & Africa Relevance

Moroccan exports of fruits, vegetables, and seafood depend on long corridors toward Europe and Sub-Saharan Africa, where avoidable cold-chain losses can directly reduce sector revenue.