Mercedes-Benz Builds Next-Generation Factories with AI and Digital Twins

Mercedes-Benz Accelerates DX in Manufacturing Through Collaboration with NVIDIA

Mercedes-Benz, a global automotive manufacturer, is digitizing its production processes to rapidly respond to increasingly complex supply chains and market demands. Central to this effort is the construction of 'digital twins' leveraging the digital production ecosystem 'MO360' and NVIDIA's 3D development platform, 'NVIDIA Omniverse'. This initiative is called the 'Digital First' approach, aiming to improve efficiency, flexibility, and intelligence by first designing and simulating in a virtual space before constructing or modifying physical factories.

Challenge to be Solved: Complex and Time-Consuming Production Line Planning

Automotive production lines are massive systems involving the complex coordination of thousands of parts, many workers, and robots. Introducing new models or modifying production equipment incurred enormous time and cost, with the risk that small errors in the planning stage could lead to significant rework. There was also an increasing need to respond quickly to supply chain fluctuations and flexibly adjust production plans. Physical trial and error had its limits, demanding a more efficient and precise planning and verification process.

How AI Was Used: Physically Accurate Simulation via Digital Twins

Mercedes-Benz is building 'digital twins' of its factories worldwide on the NVIDIA Omniverse platform. This accurately reproduces factory layouts, equipment, robots, and worker movements in a virtual space, maintaining physical accuracy. By leveraging these digital twins, AI-powered simulations are used to optimize new production line layouts and verify robot movements. This allows problems to be identified and efficiency maximized in the virtual space before making changes in the real factory. This approach is said to halve the time required for assembly hall conversions and construction, while also improving process quality.

Estimated Architecture

The following is an estimated architecture based on public information and general configurations. It does not definitively state the actual internal configuration.

• ▸"Platform Layer:" Centered around NVIDIA Omniverse, integrating data from various 3D tools in Universal Scene Description (OpenUSD) format. Integrates with Mercedes-Benz's digital production ecosystem "MO360" to utilize real-time data.
• ▸"Input Data:" CAD data for factories and equipment, supply chain information, production plan data, real-time operational data, etc.
• ▸"AI Processing & Simulation Layer:" Utilizes physically accurate simulation engines (e.g., NVIDIA PhysX) to optimize production line layouts, robot movements, and worker flow. AI models are also used for energy consumption prediction and quality control data analysis.
• ▸"Output & Utilization Layer:" Generates optimized factory layout proposals, efficient work procedures, control programs for robots, and instructions for workers. Digital twins are also shared with suppliers to strengthen collaboration.

Introduction Effects and Key Points to Watch

• ▸"Reduced Planning Time and Cost Savings:" Eliminating physical trial and error significantly shortens the time required for factory construction and line changes, reducing costs. It is expected to reduce the coordination process with suppliers by 50%.
• ▸"Improved Quality and Efficiency:" Thoroughly verifying processes in a virtual space before production begins eliminates rework and improves quality. Simulations optimize robot and worker movements, increasing production efficiency.
• ▸"Contribution to Sustainability:" At the Rastatt plant in Germany, AI was used to monitor a sub-process in the paint shop, successfully reducing energy consumption by 20%. Simulating energy efficiency on a digital twin enables sustainable factory design.
• ▸"Enhanced Flexibility and Resilience:" The system can quickly respond to unexpected changes, such as supply chain disruptions. Line reconfigurations can be rapidly simulated on the digital twin to find optimal solutions.

What Japanese Companies Can Learn

The Mercedes-Benz case offers important insights not only for the automotive industry but for Japanese manufacturing as a whole. Even if digitalizing an entire large factory at once is difficult, it's possible to start small with specific production lines or processes. Leveraging existing CAD data and identifying inefficient areas through virtual simulations to test improvement measures is an effective way to promote DX while minimizing costs and risks. The 'Digital First' mindset, which integrates the physical and digital worlds to make data-driven decisions, will be a valuable reference for companies of all sizes looking to enhance their competitiveness.

Today's AI News

Reference Links

Find Related AI Adoption Cases