A near future vision for factory of the future

A Smart Factory is a hyper flexible, adaptable production capability that can self-optimize operations across a broader network – including other factories & adjacent supply chain operations – with the ability to learn from running conditions in real or near-real time, with extension of execute design, planning, manufacturing, and delivery processes autonomously.

The foundation is design for modular operating models & systems, to compose plug-and-play manufacturing processes in meeting changing demand & customer requirements, for faster product delivery.

We will explore these 3 key capabilities in more detail:

• Being networked – very low-latency responsiveness, real-time data collection, AI and applications running on sensors.
• Self-optimized operations – self-sensing, for quality automation/in-process perception control. This is about extending AI as a service on the edge.
• Human-machine intelligence – copilot is becoming the interface for human machine cooperation. The human will always be the pilot, not the copilot, and is the human that transfers or recovers levels of autonomy how the operations are executed.

Key capabilities of factory of the future

I would like to highlight the following constituent elements of a smart factory:

• Connectivity: inter (and intra) connection, automation, and networking of assets including equipment, tools and human augmented personnel.
• Edge analytics: with AI it is possible to step towards cognitive factories – ability to learn, self-sense equipment health, self-healing, acting autonomously on quality assurance and reliability.
• Universal asset identification: imagine that you can turn on an equipment in a plug and play fashion, like you plug a peripheral to your laptop. Once switch-on, systems can identify what kind of manufacturing operations are able to perform, how that can be used in the context of the products to be produced, and how it is performing.
• Sustainability: optimize manufacturing operations to continuously decrease consumption of natural resources – energy, water, materials. For example, how to manage the trade-off between cost of energy source, carbon footprint or maximize profit selling energy back to the grid if there are self-production or energy storage capabilities.

Smart factories’ focus is not related solely to automation or intelligent machines; it is also about realizing human potential by the effect of augmentation:

• Cognitive augmentation: that improves the sensing & reasoning human ability, increases perception support, using simulation to predict or forecast results of actions taken, and prevents performance issues which refers to the slow intervention of operators when automation fails or when they are out of the loop. For example, you can use smart glasses to get equipment real time telemetry and help to identify anomalies, or you can rely on Digital Twin to predict outcomes of equipment optimization scenarios or close loop operations feedback to check operational effectiveness after commands have been given.

Shifting to autonomous, event-driven multi-agent system

The future of manufacturing includes the ability to produce tens of thousands of parts for various industries within hours, from tens of thousands of parts for a jet engine, then switch to doing medical implants. The knowledge of how to make the part lives in the software, enabled by agent driven systems.

Agent-based manufacturing is the next generation advancement that a smart factory should adopt. Using a distributed approach, each machine, a conveyor, a robot, a person, becomes an agent with AI-intelligence. Each agent is autonomous, and work together with the other agents, interacting in a collaborative way to reach a production decision orchestrated by another agent that lives in the software based on real-time data.

Agent-based manufacturing systems are pluggable systems, allowing changes in production facilities, e.g. the addition, removal and modification of equipment as well as software modules, on the fly, without the need to stop, reprogram and re-start systems. This design principle is crucial to support the current requirements imposed by customized demanding and flexibility.

The smart factory becomes a self-organized multi-agent system assisted with real-time data-based feedback and coordination.

• The Manufacturing Agent represents machines or people that perform manufacturing, testing or quality control operations.
• The Conveying Agent represents the devices (such as conveyers, robots or AGVs, people) that move the products, or components.
• The Product Agent represents products or components of the products that are processed in the plant.
• For each operation of the sequence, the Product Agent first announces it needs to be produced and negotiates in real time with the Manufacturing Agent to determine who is most capable and available of performing the operation. After being accepted for production, it negotiates with the Conveying Agent to determine an optimized route that transports the Product from the current position to the destination, such as a manufacturing cell.
• During manufacturing task execution, it can happen a quality issue, an equipment breakdown, in this case there is a renegotiation to allocate to other Manufacturing Agent or if no issues are found, it will proceed to the next manufacturing task or in the case the product is completely finished it is moved to warehouse and enters in the supply chain cycle.

Because each agent is able to communicate with the other agents they can self organize taking into consideration the nature of the operations being carried.

The key benefit of such an approach is that if production is disrupted or re-organized in some way, the same negotiation process still takes place, albeit with different machines or products making the decisions, and hence the system is relatively robust to change. On the other hand, operators are complemented with automation technologies (increasing the skills of the operators that stay in the production process) and most importantly it does not need for a long time of reprograming and reconfiguration in case there are changes in the manufacturing assets.

Constituent domains of factory of the future

Factories of the future and autonomous operations journey is a succession of evolving plateaus, rather than a linear progression towards increased maturity.

Automation, AI-driven optimization, towards autonomous operations, requires that the most fundamental enabler be real-time Data followed by Analytics & AI. A key technology is a common data model according to manufacturing standards and out of box native data connectors for manufacturing and adjacent industries living in a data fabric.

For that universal connectivity – real-time heterogeneous asset data acquisition & communication – with private 5G is required. That will allow:

• Asset Discovery & Brokering (Plug & Play) – meaning what kind of machine or equipment is, what is does, its production capacity, it is status. Hence once each asset is universally connected, it can broadcast and receive data in real time, and it can be part of an agentic system.

In an agent-driven factory, towards self-organization and self-optimization, setting a solid foundation for data & AI will allow:

• Asset operation & monitoring.
• Work order activity monitoring.
• Product plan orchestration.

As we can ingest and process data in real-time, we can start moving towards integrated factory integrity, including:

• Anomaly detection, root cause analysis, event correlation
• Autonomous troubleshooting, combining with automation.

And last, we can achieve autonomous operations with closed loop feedback, leveraging for example digital twins:

• Physical machines or digital solutions executing tasks autonomously & collaborating with humans to supplement existing processes.