Agility Robotics’ Digit performs material handling. Image credit: Accenture

Schaeffler Taps Accenture, NVIDIA, and Microsoft to Simulate Future Factories with AI and Humanoid Robots

Industrial giant Schaeffler is collaborating with Accenture, NVIDIA, and Microsoft to explore the future of manufacturing, leveraging digital twins, AI, and advanced robotics, including humanoid models. The partnership, highlighted as part of plans for Hannover Messe 2025, aims to optimize factory and warehouse operations by simulating various automation scenarios before physical implementation.

Simulating the Factory Floor

The core of the collaboration involves using NVIDIA's Omniverse platform to create detailed digital twins of Schaeffler's facilities. Accenture is leveraging its expertise to build these virtual replicas, allowing Schaeffler to simulate different layouts, workflows, and levels of automation.

This virtual proving ground serves several purposes:

1. Optimized Facility Planning: By simulating material flow and human-robot collaboration in Omniverse, Schaeffler can identify optimal layouts for production lines and kitting stations, potentially reducing commissioning times for new or retrofitted facilities.
2. Evaluating Automation Levels: The digital twins allow for testing scenarios ranging from primarily manual operations to those incorporating autonomous mobile robots (AMRs), Schaeffler's own EMMA mobile manipulator cobot, and even general-purpose humanoid robots.
3. De-risking Robot Deployment: Before deploying expensive hardware, especially complex systems like humanoids, simulations enable testing, training, and validation in a safe, virtual environment.

"As a leading motion technology company, Schaeffler leverages disruptive innovations such as physical AI, digital twins, and humanoid robots to enhance operational excellence," said Andreas Schick, COO at Schaeffler AG. Patrick Vollmer, Global Industry Group Lead for Industrials at Accenture, added that "Physical AI is reinventing industrial automation" by offering efficient ways to train and control robot fleets.

Movements are captured in the real world and translated back into Omniverse, where humanoid robots can learn them. Image credit: Accenture

Training Humanoids in the Metaverse

A key focus of the demonstration involves integrating humanoid robots into these digital twins. The partners are using Mega, an NVIDIA Omniverse Blueprint, to simulate and test fleets of robots.

Specific examples showcased include:

• Agility Robotics' Digit: Simulated performing material handling tasks like moving totes within a virtual Schaeffler facility.
• Sanctuary AI's Phoenix: Demonstrated learning real-world tasks, such as assembling spare part kits, by observing simulations within Omniverse. This employs imitation learning, potentially accelerated by capturing human movements using vision AI tools like NVIDIA Metropolis and translating them into the simulation for the robot to learn from.

Accenture's recent strategic investment in Sanctuary AI underscores the growing interest in humanoid platforms, which are seen as potentially versatile additions to human-centric workspaces.

"Tomorrow’s factories require complex collaboration between humans, industrial automation systems and multi-robot fleets," noted Mike Geyer, head of digital twins at NVIDIA, highlighting the role of Omniverse in testing these interactions at scale.

From Simulation to Real-World Optimization

The collaboration extends beyond simulation into operational optimization. Data generated from Omniverse simulations, representing various operational scenarios, is fed into Microsoft Fabric, a unified data platform.

This allows site managers to analyze Key Performance Indicators (KPIs) like robot utilization and overall equipment effectiveness (OEE) across different simulated conditions. The goal is to identify potential bottlenecks or inefficiencies and optimize the performance of physical robots on the shop floor.

Accenture and its joint venture Avanade have also developed a generative AI-powered agent for Schaeffler's Schweinfurt site. Integrated with Microsoft Fabric, this tool allows staff to query production data and troubleshoot issues using natural language, aiming to speed up problem-solving and potentially allow voice-based interaction with simulation data in the future.

While the integration of advanced simulation, AI, and humanoid robots into industrial settings is still evolving, this collaboration signals a clear direction. By bridging the virtual and physical worlds, companies like Schaeffler hope to gain the flexibility and efficiency needed for future manufacturing challenges.

Read the press release here: https://newsroom.accenture.com/news/2025/accenture-and-schaeffler-pave-the-way-for-industrial-humanoid-robots-with-nvidia-and-microsoft-technologies