Transforming Production With Digital Twins and Physical AI

Rising labor costs and workforce shortages are increasingly straining manufacturing operations. Product complexity, particularly in the automotive and industrial sector, demands more sophisticated assembly and quality control. Beyond the increased demand for accelerated simultaneous development cycles, pressures related to sustainability and regulatory compliance have also intensified.

For Schaeffler, these industry dynamics presented an opportunity to reimagine its processes with new approaches to production concept planning, physics-based process simulation, and advanced robot training to enable humanlike robots and humanoid robots in the future. 

As part of a forward looking and transformative digitalization strategy, Schaeffler aims to leverage investments in AI, digital twins, and simulation to innovate, enhance operational efficiency, accelerate time to market, and deliver advanced automation at scale. These investments, alongside its collaborations with Wandelbots, are enabling Schaeffler to establish advanced tools, processes, and automation across its global manufacturing operations.

To achieve its ambitious goals for manufacturing excellence and to effectively test and deploy advanced automation across its global network, Schaeffler built a virtual collaboration platform with NVIDIA Omniverse™ technologies and libraries. The platform leverages OpenUSD to integrate critical planning and production data from the company’s industrial expert software to enable Schaeffler teams to create physics-based digital twins for planning, simulating, and optimizing plants, machines, and workflows.

The Schaeffler platform has enabled their internal teams to drive significant speed and quality improvements in areas such as new production concepts and the training, testing, and validation of complex robotic tasks like the gripping and assembly of rare, high-value prototype parts. 

Schaeffler’s engineers use these digital twin solutions to virtually design and optimize facility layouts for both new and existing sites. They allow teams to simulate and identify optimal factory and warehouse configurations, including  the arrangement of production lines and kitting stations. By planning within a digital twin environment teams can test and optimize material flows, human-robot collaboration, and identify the optimal degree of automation for each facility. This innovative approach has dramatically increased planning efficiency and significantly reduced commissioning times for new production lines.

The development of Schaeffler’s platform was established through initial pilot phases, involving a core team of 15–20 specialists over a 12-month period. The AI computing infrastructure powering the development and deployment of the platform is currently based on a mixed architecture of local virtual machines and NVIDIA L40S GPUs.

Through integration and deployment support from its partners, including Wandelbots, the platform is being rolled out across multiple sites and use cases globally, with successful brownfield and greenfield projects already underway in China and Germany. As a result of its success to date, Schaeffler plans to integrate the platform into more than half of its global manufacturing sites by 2030.

The platform’s capabilities are demonstrated through three selected pilot projects: the assembly of electric motors, an inverter assembly line, and a Vehicle Lifetime Solutions (VLS) warehouse in Europe.