Reference architecture

Introduction

AI-PRISM will provide industrial users with human-centred artificial intelligence (AI)-based solutions to create a more efficient, resilient, digital, sustainable, and high-quality European manufacturing industry. To do so, we will develop an integrated and scalable environment with solutions adapted to dynamic and unpredictable manufacturing scenarios that require tasks that are difficult to automate and where speed and versatility are essential to meet users' needs. Furthermore, the solutions will be specific to semi-automated and collaborative manufacturing in flexible production processes and will not require specific robotic programming skills.

Our solutions ecosystem will have four main pillars:

Reference architecture

The AI-PRISM Reference Architecture (RA) defines the technical structure and establishes the relations between the key technical results identified in the DOA. This way, the AI-PRISM RA will provide a template solution for the technical implementation of AI-PRISM, through the integration of the envisioned key technical results. The objective is to provide an abstract description of the technical architecture of AI-PRISM to identify the main equipment and communication networks involved, and how the different AI-PRISM components interact between them within this technical infrastructure

First, it establishes the relationships between hardware components and communication networks, runtime environments, and software components. Then, for each of the software components, this document provides a description, a list of functional components mapped to the AI-PRISM RA and an indication of the expected input and output interfaces. Also, the main interactions between components will be identified, as well as interactions with functions outside the scope of AI-PRISM.

The AI-PRISM RA is based on standard based reference architectures and reference models. The AI-PRISM RA is based on the IIRA three-tier architecture pattern, which is a multi-tiered architecture pattern which comprises the edge, platform, and enterprise tiers, and three networks, the proximity network, the access network, and the service network. These concepts are adapted to the AI-PRISM vision in the AI-PRISM RA, which divides the system into 3 primary zones or tiers, namely the device tier, the fog tier, and the device tier.

Reference Architecture

Reference Architecture Components

The following table contains links to the detailed specifications of each component in the reference architecture.

Acronym Name Description Link
RF AI-PRISM ROS Framework ROS Framework to manage the deployment of ROS modules. RF Documentation
BH AI-PRISM Base Hardware Supported base hardware interfaces (for AGVs, movile robots, industrial robots). BH Documentation
CM AI-PRISM Communications Modules ROS packages that will enable real time communication within the AI-PRISM platform and with other devices. CM Documentation
AS AI-PRISM Ambient sensing infrastructure Supported sensors and cameras to sense the environment. AS Documentation
RC AI-PRISM Real Time Communications Network Supported real time communication hardware. SDN, QSDN, WSN sensor network control and management. RC Documentation
IP AI-PRISM IIoT Platform Edge/Cloud platform to support the deployment of data services, simulation services and ROS Framework. IP Documentation
DS AI-PRISM Data Platform Data services to support AI based solutions (AI-PRISM Data Model, the Data Broker, the Repositories, the Data Management functions). DS Documentation
SE AI-PRISM Simulation Environment Development of simulation tools representing production plants, production lines, and human agents. SE Documentation
AD AI-PRISM Ambient Digitalisation Modules Comprehends software modules to integrate data acquired with the AI-PRISM Ambient sensing infrastructure. AD Documentation
CD AI-PRISM CI/CD Framework for AI-based Solutions Pipelines and infrastructure to support AI based development, integration and validation. CD Documentation
PE AI-PRISM AI-based Perception Enhancing Modules AI-based modules to enhance perception capabilities. PE Documentation
DR AI-PRISM AI-based Agent Level Reasoning Enhancing Modules An AI-based module that handles the discrete level reasoning/planning between a single robot system and a user interacting with the system. DR Documentation
CR AI-PRISM AI-based Ambient Level Reasoning Enhancing Modules AI-based modules to enhance collaboration of agents in the ambient. CR documentation
HI AI-PRISM Human - Machine Interaction (HMI) Modules Integrate the sensor-modalities used in the project in a probabilistic motion/force model. HI Documentation
PD AI-PRISM Programming by Demonstration Environment Integration of the constraint-based framework eTaSL in the project middleware. Development of the appropriate drivers for the robot system. PD Documentation
SP AI-PRISM Human Safety Management Procedures Monitoring and managing human safety in workplaces. SP Documentation
NS AI-PRISM Network Suite Service suite composed of a dynamic web-based user interface, an online Virtual Pilot builder, user access control, digital certification, interactive training engine, scheduling, and optimization engine. NS Documentation

Functional Specifications

You can find detailed functional specifications in this link.

Technical Specifications

You can find detailed technical specifications in this link.

Usage Specifications

You can find detailed usage specifications in this link.

Developers guide

If you are a developer and want to contribute to the documentation or the code, please read the Developers Guide.

Support

tomasz.kolcon@piap.lukasiewicz.gov.pl

Roadmap

After completion of work in the project or specific tasks, selected results will be published as open-source.

Contributing

Changes should be made in a new branch and create a mere request, unless otherwise stated.

Authors and acknowledgment

AI-PRISM team.

License

TBD.

Project status

Under development