Applying Digital Twin and AMR Technologies for Industrial Manufacturing
View the Infographic Read the Report Check out the code on githubMon, 29 Apr 2024 15:35:04 -0000
|Read Time: 0 minutes
Globally, there are millions of factories and industrial facilities running around the clock, often with highly complex processes in potentially dangerous environments. By leveraging Digital Twin and AMR technology, organizations can further optimize and monitor these processes, resulting in safer factory environments, less downtime, and greater quality control.
Traditionally, complete physical replicas of manufacturing facilities or mockups of industrial environments are created to test failure scenarios which are impossible to test in a production environment. These physically accurate and photoreal simulations of the actual environment are referred to as ‘Digital Twins.’
Digital Twin, AI, and Autonomous Mobile Robot(s), or AMR, technology has the potential to reduce cost by eliminating the need to create these ‘analog twins’ and instead model these environments in a digital space. Additional functionality, including detecting anomalous or hazardous conditions within factory and industrial environments, such as fires, spills, or broken machinery using AMRs, is possible. However, to detect hazardous conditions, defects, or safety events, a complete Digital Twin ecosystem that includes an array of hardware, software components, and models needs to be brought to bear.
The Proof of Concept detailed here serves as a specific use case that explores a theoretical practical deployment of these concepts and technologies as applied to an Oil and Gas Industrial Environment. Also provided is a github repo that allows for exploration and experimentation with the components and code developed for this PoC.
This challenge introduces an ideal use case for combining AI with Metaverse technology. By leveraging Digital Twin factory environments with the Omniverse, organizations can train on synthetic data for object or hazard detection and test equipment failures and hazardous events while maintaining accuracy in their specific environments. These trained and validated solutions can then be deployed in physical factory settings.
This PoC serves as a use case of a Dell Validated design, which has been created to clarify and provide clear context and a conceptual framework for understanding the building blocks of Digital Twins and their use cases. This DVD provides a rigorous technical explanation of each discreet technology behind a deployment of an Ominverse powered Digital Twin. This DVD is available on the Dell Design Hub here: https://infohub.delltechnologies.com/en-us/t/digital-twin-journey-computer-vision-ai-model-enhancement-with-dell-technologies-solutions-nvidia-omniverse/
The following PoC solution demonstrates one such strategy for a readily deployable AI-powered manufacturing solution developed using a digital twin Metaverse.
Solution Brief
Scalers AI, in collaboration with Dell and Broadcom, has showcased an AI-driven application designed for the manufacturing industry. The proof-of-concept developed by this partnership aims to identify risks in an industrial environment and offer predictive upkeep for factory machinery. The solution is particularly tailored to spot chemical leaks on the factory premises and detect bearing anomalies in industrial compressors. An overview of the PoC solution is as follows:
- An offshore Oil and Gas Rig digital twin is created with NVIDIATM Omniverse. Simulated chemical spills are distributed across this virtual offshore Oil and Gas environment.
- Virtual Autonomous Mobile Robots (AMRs) are strategically placed across the factory floor in the simulation. These robots stream video data to an AI-powered image detection system that has been trained to identify chemical spills.
- Data from sensors on virtual compressors, positioned in the Metaverse, is gathered periodically and shared via the OPC Unified Architecture (OPC UA). An analytics module, driven by machine learning, processes this data to identify any bearing defects.
- A visualization dashboard provides monitoring and alerting of both chemical spill detection and compressor failure. The dashboard includes direct views from AMR cameras, a map of the factory floor with live AMR locations, chemical spill incident logs, and a time series graph of compressor failures. Alerts of hazards are displayed on the dashboard and can additionally be sent as mobile notifications
A high-level overview of the PoC solution can be found in Figure 3.
Figure 3: Solution Overview (Source: Scalers AI)
The key hardware components used in the solution included the following:
Key Takeaways
This PoC solution illustrates novel and valuable technologies for AI practitioners and IT operations and decision makers working in industrial and manufacturing environments. Notable highlights from the solution include:
- Showcasing a feasible AI solution for industrial settings, made possible by Dell PowerEdge and utilizing NVIDIA Omniverse for the generation of training data and testing of the implementation. The Metaverse solution can be adapted to a real-world factory scenario.
- The capabilities for detecting equipment failures and predictive maintenance, powered by AI, are facilitated using data from machine sensors and the OPC UA protocol, which is frequently used in industrial settings.
- The proof-of-concept takes advantage of the GPU-rich Dell R760xa servers equipped with NVIDIA L40S GPUs. These servers facilitate the simulation of the Metaverse, video processing, and AI inferencing. A modular architecture connects the AI pipeline and visualization dashboard over high-bandwidth Broadcom Ethernet for a highly scalable solution.
Conclusion
The application of AI technology in industrial settings offers substantial potential to streamline operations, improve safety measures, and minimize expensive operational interruptions. However, despite the promising benefits, manufacturing firms face numerous hurdles when implementing new AI solutions. These challenges span from comprehending the hardware prerequisites and integrating the solution with existing systems to feasibly creating and testing solutions without causing disruptions in active factory environments. This proof-of-concept illustrates how these obstacles can be overcome by harnessing the innovative Metaverse technology in conjunction with robust hardware from Dell, NVIDIA, and Broadcom.