AI-Enhanced Digital Twin for Conveyor Systems and Production Planning

Use Case:

Deployment of an AI-supported digital twin and declarative planner to model simulate and optimize material flow in complex conveyor-based assembly lines enhancing OEE and adaptability.

Outcome:

• Achieved TRL 7 for the digital twin with highly accurate synchronization to real-world sensor data;
• Developed and validated a planner using PDDL and Julia Planners toolset;
• Enabled realistic planning under normal and degraded scenarios (e.g., conveyor malfunction);
• Significant progress toward online planning, including execution on distributed nodes;
• Created graphical editor prototype for scalable twin-model generation.

Ecosystem Support:

StairwAI-provided expert matchmaking, funding flexibility for AI consulting, and mentoring from UNIBO and industrial partners including Bosch Bühl

AI Relevance:

The project demonstrates how SME-led digital transformation can benefit from low-barrier AI tools and domain-aligned declarative planning, emphasizing:

1. AI-enhanced industrial planning without heavy HPC needs;
2. Interoperability with existing PLCs and sensor systems;
3. Real-time coordination between digital simulation and physical systems.

Summary:

Peer Stritzinger GmbH successfully advanced the development of a Digital Twin-Coupled Industrial Transport Planner during the StairwAI support program. The goal: optimize conveyor-based production lines using a high-fidelity digital twin connected to a PLC and integrated with an AI planner. The twin was developed with direct factory floor integration (via OPC-UA), achieving TRL 7 accuracy in synchronizing simulation with real machine states—even with limited sensor data. For AI-based planning, the team tackled a key challenge: finding an expert with domain-specific planning knowledge. Through a bespoke StairwAI survey, they identified a capable AI planning expert who helped model the conveyor system using PDDL and deploy the Julia Planners library. The planner supported fault-resilient operation and online plan generation. Notably, the planning domain was structured to maintain FIFO ordering and enabled distributed plan execution.

The team also began developing a graphical editor for easier future expansion of digital twins. While the planner component reached TRL 4, integration work continues beyond the project. Their roadmap includes autogenerated PDDL from twin models, realistic OEE-based optimization, and distributed Erlang deployment for Plug&Produce configurations. The business model centers around Green and Brown Field applications: low-cost retrofitting or new smart assembly systems. Financially, benefits include measurable OEE gains, SaaS-based digital twin tooling, and a scalable edge-to-cloud control stack. Peer Stritzinger GmbH’s journey highlights the StairwAI ecosystem’s role in transforming advanced AI concepts into practical, deployable manufacturing solutions.