Accenture has launched “Physical AI Orchestrator,” a solution to help manufacturers reinvent existing and future factories and warehouses to become software-defined. The cloud-based solution combines NVIDIA Omniverse, including the “Mega” NVIDIA Omniverse Blueprint, NVIDIA Metropolis and AI agents from Accenture’s AI Refinery™ platform.
In a software-defined facility, virtual replicas mirror the physical automated plant or warehouse and its equipment. These live digital twins detect issues and use accurate physics to simulate the impact of potential process changes in real time. AI Agents then convert the insights into precise instructions, which enable the physical plant to adapt to changing demand, quality or scheduling. Accenture’s Physical AI Orchestrator lets manufacturers build live digital twins of planned and existing physical assets, from conveyors and industrial and mobile robots to shop floor and warehouse layouts, and connect them to physical counterparts.
For example, Belden, a network and data solutions provider, has developed a solution for worker safety in factories and warehouses. Using Physical AI Orchestrator, it developed a virtual safety fence solution to create safety zones around robots without disrupting ongoing operations. If a human enters the zone, the robots are automatically stopped or rerouted. The solution uses edge AI to detect and model worker, vehicle and robot movements and equipment paths at centimeter-level fidelity within facilities. Trained on multiple what-if scenarios, such as forklifts reversing unexpectedly, the virtual safety fence is expected to be first deployed by an automotive manufacturer to address pedestrian safety in warehouse environments.
Other examples of companies taking advantage of Physical AI Orchestrator include:
- A life sciences company can now validate ideal production conditions for biologics and vaccines faster than if using physical equipment. Using Physical AI Orchestrator to simulate the entire preservation cycle and drying process, the company tested various temperature and pressure profiles. The simulation showed exactly when, where, and why deviations occurred. This resulted in reduced batch variability and increased product shelf life.
- A consumer goods manufacturer has increased warehouse throughput and reduced capital expenditure. Using Physical AI Orchestrator, it built a digital twin of its warehouse operations to analyze worker movement, picking rates and conveyor systems to identify throughput gaps and layout inefficiencies. The twin recommended adjustments to the layout design and resource allocation. This resulted in a 20% throughput improvement from optimizing the conveyor flow and saved 15% in capital expenditure by eliminating iterative trial-and-error redesign.
“Physical AI Orchestrator acts as a brain for a physical space,” said Prasad Satyavolu, Americas Lead of Accenture’s Digital Engineering and Manufacturing Service, Industry X. “Powered by NVIDIA Omniverse technologies and Accenture AI Refinery, it is designed to enable software-defined factories and to make agentic AI and physical AI part of the fabric of manufacturing. We are already seeing it provide quick and lasting benefits to our clients across the globe. This is particularly relevant to companies in the US, where manufacturing reinvention is a prerequisite for reindustrialization.”
Early adopters have seen measurable benefits with Physical AI Orchestrator, which is designed to significantly lower product design costs and capital costs for building new plants and produce higher factory planning accuracy.
The capabilities Physical AI Orchestrator offers to manufacturers include but aren’t limited to the following:
- Reality capture – a set of automated, AI-powered processes to generate digital twins of factories and equipment. Videos and scans are converted into photorealistic 3D models to detect layout changes and keep the digital twin updated.
- AI agents – aiding engineers during the entire process of, for example, designing, simulating and installing a new production line, with expert knowledge from past projects.
- Vision analytics – to capture and ingest live video data, such as worker, vehicle and material movements in a facility required to run precise simulations.
- Asset connectors – to integrate existing equipment into a simulation in a plug and play fashion.
- XR extensions – enabling interactions with the digital twin in a high-quality augmented environment and training and collaboration in a shared virtual space.
