Enterprise & Industry

Inside the AI Warehouse: How Otto Group Is Teaching Robots to Work Together

Using Nvidia's Omniverse, the retail giant created a digital twin to orchestrate robots from different vendors.

Deep Dive

At Nvidia's GTC, the Otto Group revealed how it's solving a major logistics hurdle: getting robots from different vendors to work together. The German retail giant, which serves 45M customers and manages its entire supply chain, has warehouses filled with disparate robotic systems—autonomous mobile robots (AMRs), shuttle systems, and robotic arms. Instead of replacing them, Otto built a Coordinated Autonomy Layer (CAL), a software layer that sits between the warehouse management system and the machines, assigning tasks and managing movement across the floor to prevent collisions.

The project began by creating a precise digital twin. Using a Boston Dynamics Spot robot to scan a pilot warehouse, Otto collected data over a week to build a 3D model in Nvidia's Omniverse platform. This digital twin, more accurate than their own blueprints, allowed them to simulate and optimize material flow before implementing changes on the physical floor. One simulation-led layout change reduced robot stop-and-go events by 20%, directly boosting productivity. The next phase involves adding an 'artificial brain' AI layer on top of CAL to dynamically adjust task assignments and routing based on real-time data, all validated within the Omniverse simulation.

Key Points
  • Built a Coordinated Autonomy Layer (CAL) to unify robots from different vendors, acting as middleware between warehouse management and machines.
  • Created a precise 3D digital twin in Nvidia Omniverse using a Boston Dynamics Spot robot, enabling layout simulations that cut robot stop-and-go events by 20%.
  • Planning an 'artificial brain' AI layer for dynamic, data-driven task assignment, moving toward a self-controlled warehouse under human supervision.

Why It Matters

This proves AI can orchestrate complex, real-world physical systems, turning fragmented automation into cohesive, efficient fleets that boost productivity.