Audi Győr MDB — production-line logistics system

Executive summary

A new high-output engine and vehicle assembly line at Audi's Győr plant needed internal logistics capable of accurate material supply, minimal human intervention and continuous traceability. The starting arrangement relied on paper-based picking and manual intervention, making supply slow, error-prone and risky for production rhythm.

Rather than partially automate existing logistics, the team effectively redesigned the process. Sensors monitored component stock on line-side racks. Falling below a defined minimum sent a central signal; the system dispatched the correct AGV with a prepared picking trolley to the required rack. Full parts containers moved onto the rack and empty containers returned by gravity without human assistance. FIFO storage, gravity roller tracks and RFID identification completed the design.

The original concept included a fully automated high-bay warehouse, which Audi rejected on cost and strategic grounds. The final design deliberately divided work between people and machines: trained staff loaded trolleys, while line-side transport, demand detection and much of the delivery cycle were automated. The overall system reached approximately 90% automation.

The source lists the following results:

• error-free operation on the first live test day;
• elimination of logistics-related line stoppages;
• elimination of previously occurring incorrect-component deliveries;
• a substantial reduction in direct line-side logistics staffing requirements;
• and, alongside improved ergonomics, sickness absence falling to 30% of the previous production line's level.

Project profile

1. Mirror — initial operations and the real risk

1.1. Paper-based picking beside a high-output production line

The source describes a paper-based picking structure with substantial manual intervention. This is not inherently wrong: paper and human coordination can work in simpler, slower environments. Beside a high-output automotive production line, however, the same arrangement falls into a different risk category.

The question is not whether parts are generally available, but whether the right part arrives in the right quantity, at the right time and at the right line-side point. Failure in any of these four conditions can disrupt production. Stable material movement therefore requires an equally stable information chain.

1.2. A line stoppage is a production event, not a logistics inconvenience

A line stoppage could present a loss risk of several million euros per day. Earlier logistics-related interruptions could represent several million euros in annual losses. These statements use different time horizons; this study neither combines them nor derives a new financial figure.

Internal logistics is critical production infrastructure here. Uncertain logistics makes production availability uncertain. The objective therefore went beyond faster material movement: it also needed to reduce errors, information delays and incorrect decisions.

1.3. The wrong component as an information and quality failure

The source specifically reports that digitalisation eliminated earlier errors, including incorrect components arriving for an engine. Material-supply failures can therefore become direct quality-assurance issues, rather than remaining warehouse or transport problems.

Physical objects and their information must stay connected. Parts, containers, picking trolleys and AGVs need identification, and the system must know which production requirement they serve. RFID was a means of joining physical and digital states, rather than simply a modern technology choice.

2. The causal and information map of the problem

The source is a historical project description, not a comprehensive forensic audit. The following map separates documented facts from today's Kyntrax interpretation, particularly when drawing modern methodological lessons from an earlier project.

2.1. Where was the real control point?

Monitoring line-side stock was central. A component falling below its minimum shelf level became a control event rather than merely a warehouse data point. The sensor signal triggered central control, linking the correct AGV and prepared trolley to a specific production requirement.

Physical reality directly generated an execution task. There was no need to wait for someone to notice depletion, phone the warehouse, complete paperwork and initiate transport manually. Both decision delay and opportunities for misunderstanding decreased.

3. Compass — which target state was needed?

The source describes the objective as fully automated, self-propelled supply with minimal staffing, aligned with Audi's production rhythm. The target nevertheless comprised several interdependent operational requirements, none meaningful in isolation.

3.1. Demand-driven material supply

Transport responded to actual line-side consumption rather than fixed dispatch times. Scheduled supply assumes sufficiently predictable consumption and production rhythms. Demand-driven supply triggers the next action from actual stock conditions.

3.2. Minimal human intervention — but not at any cost

The original design included an automated high-bay warehouse with robotic picking. Audi did not implement this for cost and strategic reasons. The team adapted rather than defending the original plan for technological prestige. Trained staff continued loading trolleys while the delivery cycle was automated.

The goal is an economically and operationally justified automation boundary, rather than the highest percentage. A better system allocates each task to the appropriate participant under suitable controls; it does not require machines everywhere.

3.3. Traceability as an operational requirement

The source states that RFID identified every container and AGV, allowing real-time tracking of components within the system. In automotive operations, traceability supports quality assurance, discrepancy investigation and responsibility, beyond logistics convenience.

3.4. FIFO as implemented physical logic

Gravity roller tracks supported FIFO. The rule became part of the physical arrangement, helping the first stored unit become the first used. Software, mechanical design and process logic worked together.

4. Tunnel — redesigning the entire material supply cycle

4.1. The process step by step

1. Warehouse staff prepare the required components on picking trolleys according to the defined workflow.
2. Sensors monitor stock levels on line-side supply racks.
3. When a component drops below its defined minimum, a signal goes to central control.
4. Control dispatches the AGV carrying the appropriate prepared trolley to the correct production line.
5. The AGV positions itself at the line-side rack system.
6. Full parts containers are transferred onto the supply rack.
7. Empty containers return to the AGV at the same time.
8. The AGV returns empty containers to the warehouse for refilling or recirculation.
9. RFID identification and system state keep objects and movements traceable.

4.2. What changed from the previous process?

The largest change was the structure of the decision chain, not merely replacing a driven forklift with an AGV. Previously, people had to notice, communicate, interpret and execute demand at several points. Now physical conditions became direct, controlled system events.

5. Technical components and their business meaning

5.1. AGV: controlled execution, not movement alone

An AGV by itself is a material-handling vehicle. Its value here came from a specific task derived from the production-line state. Route, load and destination belonged to the process, rather than isolated vehicle movement.

5.2. Sensor: from measurement to business event

Shelf-level sensor data acquired business meaning: crossing the minimum threshold created a replenishment requirement. This is where digitalisation becomes operational intelligence: understanding what data means for the next action, rather than merely collecting it.

5.3. RFID: connecting the physical and digital worlds

RFID identified AGVs and containers, supporting real-time visibility and traceability. Physical objects gained digital identities, and the system knew their process states. That relationship mattered more than RFID technology alone.

5.4. FIFO and gravity tracks: building rules into the physical system

FIFO is a familiar quality and stock-rotation principle. Reliance solely on human discipline permits exceptions and errors. Gravity roller tracks embedded the rule in the physical environment, strengthening control through design rather than added administration.

5.5. Container recirculation: plan the end of the process too

Delivering a full container is only half the process. Empty-container return prevents congestion, container shortages and warehouse disruption. The source describes AGVs returning empties too, showing a closed cycle rather than one-way transport design.

6. Visual evidence of the implemented system

The original historical study includes two photographs of the implemented production logistics environment. They illustrate the physical layout and described components, but do not replace full technical handover documentation.

The image shows segregated, unobstructed travel lanes and modular supply structures. Reliable autonomous vehicle operation requires physical space, traffic rules and process logic to be aligned.

The second image shows gravity tracks and container handling. Physical design supports FIFO while unique identifiers enable digital tracking, combining physical control with information-system traceability.

7. People and automation: why 90% matters, and why 100% is unnecessary

The source reports approximately 90% automation. How this was achieved matters more than the percentage. After the customer rejected the automated high-bay warehouse and robotic picking on cost and strategic grounds, trolley loading remained human work while line-side transport and much of the control were automated.

7.1. The right automation boundary

Automating everything can create excessive investment, rigidity and complexity that is difficult to justify economically. Ask where automation materially reduces errors, delay, risk or cost, rather than which work can be mechanised.

The critical issue was predictable replenishment. Human picking could remain within a controlled, trained workflow. Transport, stock monitoring and execution signals were repetitive, time-critical tasks where automation directly improved stability.

7.2. Human work within an automated system

Automation did not eliminate the human role. People remained where flexibility, visual checking and adaptability added value; the system took over repetitive movement, precise timing and consistent rule-following.

8. Commissioning, evidence and results

8.1. First live test

The source reports error-free operation on the first live test day: AGVs followed designated routes with correct timing, and storage and supply worked smoothly. No detailed test record, KPI time series or failure statistics accompany this strong historical claim. It is reported as a source statement, not converted into independently audited technical evidence.

8.2. Production-line availability

The historical account reports no production losses caused by logistics faults after implementation, and elimination of incorrect-component deliveries. These are among the strongest business outcomes, directly linking logistics process quality to production stability.

8.3. Staffing and ergonomics

The source reports a substantial reduction in required line-side logistics staff. With ergonomic line supply, sickness absence fell to 30% of the previous line's level. No detailed HR dataset or controlled study is provided, so this remains a documented project claim rather than a generalisable causal research result.

8.4. Strength of evidence

9. Lessons for manufacturing businesses today

9.1. Internal logistics is part of the production system

Production and logistics may be separate organisational functions, but belong to the same value-creation chain. If logistics cannot supply materials on time, production cannot meet its objectives.

9.2. Design event logic before automation

An AGV, robot or sensor cannot define good operations alone. First specify which states mean which business events, which thresholds trigger tasks, how the right material and destination are selected, and which confirmation closes the action.

9.3. Design physical and information processes together

Mechanical equipment, sensors and digital identification are not separate projects. Gravity tracks, RFID and AGVs support the same business objective: reliable, traceable delivery of the right material to the right location.

9.4. Exception handling matters as much as normal operation

The source mainly describes normal operations. A modern implementation would separately address missing trolleys, incorrect RFID identification, failed AGVs or sensors, unexpected priority changes, blocked routes and urgent manual intervention. High automation produces stability only when exceptions also have predefined operating rules.

9.5. Good automation can reduce bureaucracy quietly

Control improved without additional forms. Sensors, RFID and machine movement automatically generated event data that previously might have required separate human administration. This is an important meeting point of control and efficiency.

10. What would this mean today in Industry 4.0 and enterprise AI?

The historical logic remains current: physical state creates an event, the event gains business meaning, a rule drives a decision, execution creates a new state. Enterprise AI does not replace that chain. It can safely add value when built on this structured operational foundation.

10.1. Where could AI add value today?

• Forecasting component consumption and replenishment workload from production plans, actual consumption and historical patterns.
• Recognising abnormal AGV movement, stock consumption or route loads and recommending investigation.
• Connecting logistics events to quality deviations to identify recurring root causes earlier.
• Using simulation to recommend minimum-level or route changes, with human or deterministic approval before execution.
• Using maintenance and incident data to anticipate risks involving AGVs, sensors or supply equipment.

10.2. What should not be delegated without context?

AI should not freely change minimum stock, production priorities or AGV routes merely because it can. It must know business rules, safety constraints, quality requirements and authorisation boundaries. Capability and authority are distinct in Kyntrax's AI approach.

10.3. Why did this project reflect AI-readiness thinking before AI?

It created operational states, identified objects, events, thresholds, rules and feedback without using AI. Without such foundations, later AI would see documents and databases without reliably understanding their operational meaning.

11. The project's place in Kyntrax's methodological heritage

Mirror–Compass–Tunnel was not yet formalised under today's terminology during the project. The same progression of thinking is nevertheless recognisable in the historical work and is now organised into Kyntrax's explicit methodology.

11.1. Why is this relevant to Kyntrax?

The project demonstrates Kyntrax's principle of selecting technology after understanding actual operations. AGVs, RFID and sensors answered a precisely defined operational problem rather than forming the starting point.

Operational improvement also extended beyond software. Racks, roller tracks, routes, picking workflows and digital control formed one system. This informed Kyntrax's later principle of viewing a company as connected operational events rather than a collection of modules.

12. Frequently asked questions

What was the main problem in the Audi Győr MDB project?

The initial supply process relied on paper-based picking and substantial manual intervention. It was slow, error-prone and risky for continuous supply to a high-output production line.

What triggered replenishment?

Sensors monitored line-side shelf stock. When a component dropped below its minimum, central control dispatched the correct AGV with the appropriate picking trolley.

Which technologies did the system use?

The historical account describes AGVs, stock-level sensors, FIFO gravity roller tracks, RFID identification and integrated container handling.

Was the system fully automated?

No. The planned automated high-bay warehouse was not implemented, so people continued loading picking trolleys. The overall logistics system nevertheless reached approximately 90% automation.

Why does FIFO matter in this project?

Gravity roller tracks physically supported first-in, first-out use, so FIFO was more than an administrative instruction.

What was RFID's role?

RFID identified containers and AGVs and digitally tracked physical movements, supporting quality assurance and discrepancy investigation.

Which outcomes does the source document?

An error-free first live test, elimination of logistics-related stoppages and incorrect-component deliveries, reduced direct logistics staffing and significant ergonomic improvement.

Why is this relevant to AI projects today?

Even without AI, the system created structured context connecting objects, states, events, thresholds, rules and feedback. Enterprise AI needs such context to operate reliably.

What is Kyntrax's main lesson from this case?

Begin by understanding the business event to control, the information behind decisions and the execution chain ensuring the result. Select technology afterwards.

13. Concept map — a quick guide for people and machines

14. Sources, evidence and publication context

14.1. Primary source

The primary historical source is the “Audi Győr MDB Production Line Logistics System” chapter of “Case Studies (4) EN”, held by Kyntrax. This English page adapts the approved Hungarian analysis, which organises and explains the historical project data through today's Kyntrax operational diagnostic approach.

14.2. What comes directly from the source?

• The description of initial paper-based, manual logistics.
• The automated target state with minimal human intervention.
• Use of AGVs, stock-level sensors, FIFO, RFID and container handling.
• Rejection of the automated high-bay warehouse and retention of human picking.
• The 90% automation claim.
• The error-free first live test claim.
• The reported elimination of logistics stoppages and incorrect-component deliveries.
• Historical sickness-absence reduction associated with ergonomic design.
• The two photographs of the implemented environment.

14.3. What is Kyntrax interpretation?

The causal map, information-flow analysis, Mirror–Compass–Tunnel mapping, modern Industry 4.0 and AI interpretation, authorisation boundaries and implementation lessons are current professional interpretations grounded in historical facts. They are not presented as verbatim statements from the original project documentation.

14.4. What the source does not include

The available source contains no complete technical design, investment costs, detailed ROI calculation, full KPI time series, formal test report, detailed HR statistics or dataset from which additional financial or productivity metrics could be reliably calculated. This study does not invent them retrospectively.

14.5. Author and methodological background

Joseph Bakus — Founder & System Architect, Kyntrax. Creator of the Kyntrax operational methodology and system architecture. This document analyses the earlier industrial project through the Kyntrax approach.

Related Kyntrax topics

• Business diagnostics before digitalisation
• Uncovering information flow blind spots
• Operational audit before AI implementation
• Process assessment before ERP replacement
• Mirror — uncovering real operations
• Compass — defining the target state and priorities
• Tunnel — controlled implementation and change management

© 2026 CLOUD DEVELOPMENT AND SERVICES LTD · Kyntrax. All rights reserved.

The content may be used for search, citation, summaries and short quotations with clear Kyntrax attribution. Commercial reuse and methodological adaptation require separate permission.

Terms of use and intellectual property →