Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

An Agentic Framework for Rapid Deployment of Edge AI Solutions in Industry 5.0

Jorge Martinez-Gil, Mario Pichler, Nefeli Bountouni, Sotiris Koussouris, Marielena Márquez Barreiro, Sergio Gusmeroli

TL;DR

The paper addresses rapid deployment of Edge AI in Industry 5.0 by proposing a modular, agentic framework that performs local inference on edge devices using a suite of components (Config Loader, Inference, UI, GenAI, Designer, and Design). It leverages MQTT and REST-based integrations to enable real-time data ingestion, prediction, explanation, and automated deployment of updated pipelines, with HITL support. A use case in the food-production domain demonstrates improved deployment speed, low end-to-end latency (under 200 ms), and high predictive accuracy, showcasing the framework’s practicality and adaptability. The work highlights the potential of combining Collaborative Intelligence and agent-based task allocation to deliver privacy-preserving, human-centric edge AI across diverse industrial settings, while outlining future directions for efficiency and broader domain validation.

Abstract

We present a novel framework for Industry 5.0 that simplifies the deployment of AI models on edge devices in various industrial settings. The design reduces latency and avoids external data transfer by enabling local inference and real-time processing. Our implementation is agent-based, which means that individual agents, whether human, algorithmic, or collaborative, are responsible for well-defined tasks, enabling flexibility and simplifying integration. Moreover, our framework supports modular integration and maintains low resource requirements. Preliminary evaluations concerning the food industry in real scenarios indicate improved deployment time and system adaptability performance. The source code is publicly available at https://github.com/AI-REDGIO-5-0/ci-component.

An Agentic Framework for Rapid Deployment of Edge AI Solutions in Industry 5.0

TL;DR

The paper addresses rapid deployment of Edge AI in Industry 5.0 by proposing a modular, agentic framework that performs local inference on edge devices using a suite of components (Config Loader, Inference, UI, GenAI, Designer, and Design). It leverages MQTT and REST-based integrations to enable real-time data ingestion, prediction, explanation, and automated deployment of updated pipelines, with HITL support. A use case in the food-production domain demonstrates improved deployment speed, low end-to-end latency (under 200 ms), and high predictive accuracy, showcasing the framework’s practicality and adaptability. The work highlights the potential of combining Collaborative Intelligence and agent-based task allocation to deliver privacy-preserving, human-centric edge AI across diverse industrial settings, while outlining future directions for efficiency and broader domain validation.

Abstract

We present a novel framework for Industry 5.0 that simplifies the deployment of AI models on edge devices in various industrial settings. The design reduces latency and avoids external data transfer by enabling local inference and real-time processing. Our implementation is agent-based, which means that individual agents, whether human, algorithmic, or collaborative, are responsible for well-defined tasks, enabling flexibility and simplifying integration. Moreover, our framework supports modular integration and maintains low resource requirements. Preliminary evaluations concerning the food industry in real scenarios indicate improved deployment time and system adaptability performance. The source code is publicly available at https://github.com/AI-REDGIO-5-0/ci-component.

Paper Structure

This paper contains 17 sections, 4 figures.

Table of Contents

  1. Introduction
  2. State-of-the-Art
  3. Antecedents
  4. The Present
  5. Open Challenges
  6. Contribution Over the State-of-the-art
  7. A Framework for Rapid Deployment of Edge AI Solutions
  8. Architecture
  9. Initialization
  10. UI Agent
  11. GenAI Agent
  12. Inference Component
  13. Design Component
  14. Components Working Together
  15. Use Case
  16. ...and 2 more sections

Figures (4)

  • Figure 1: Our framework begins when the Config Loader initializes both CSV Reader and Sensor Streaming, routes data through the MQTT Broker to the Inference component, returns predictions to the UI Agent, invokes the GenAI Agent via ChatGPT4o for on-demand analysis, and uses the Design component to deploy updates to the inference component
  • Figure 2: Start screen available to the operator to initialize each of the components of the framework
  • Figure 3: Browser-based screenshot of the CI component with the functionality that allows monitoring and correction of the models hosted on the Edge
  • Figure 4: Data analysis features of the CI component, showing updated graphs next to controls in a split-view layout