Industry 4.0 means smarter manufacturing. By integrating AI, IoT, analytics, and other connected technologies into factory operations, manufacturers can initiate a new wave of efficiency without major investments.
IBM’s own experience as a manufacturer of mainframes, servers, storage, and IT appliances provides a good example. We’ve learned first-hand how using AI-powered visual inspection of components and assemblies can significantly improve process efficiencies and product quality.
As electronics components have become more complex and systems gotten denser, manual inspection is more difficult. Technicians working at a station can become fatigued inspecting complex assemblies. This can lead to defects being missed or caught at later stages when repair costs are higher.
Traditional automated inspection systems, generally rules-based, aim to catch all defects, but they often generate false positive calls. These then require extra manual inspections. In addition, as technology becomes denser, manufacturers may have to replace outdated automated inspection equipment to meet the needs of new products.
IBM uses both manual and automated inspection. In 2018 we complemented these with IBM Maximo Visual Inspection, an AI-powered computer vision solution. Running on a Power Systems or x86 GPU server, the solution applies deep learning models to automatically detect quality defects in manufacturing. A big plus is the solution’s ease of use—our manufacturing engineers and technicians can train models without help from AI experts or data scientists.
We integrated Visual Inspection into our production lines at manufacturing sites in Canada, Hungary, Mexico and the US. Whether inspection is initiated by the operator or automatically through a programmable logic controller, deep learning recognizes defects at scale with high accuracy. We’ve realized up to 5x efficiency gains and a 20% reduction in false positives, helping us maintain the highest levels of quality and production throughput. And it’s not a “set it and forget it” solution, new use cases continue to surface throughout our operations.
Our experience is a plus for clients. Those that need computer vision get help from Visual Inspection experts who have implemented the solution at our plants. Their real-world expertise helps clients achieve fast time to value.
The deep learning model, specific to each use case, typically takes just hours to train by analyzing known images and videos. After connection to a camera, the model is ready for production and can be run on a variety of systems or from the cloud. Analysis of operator feedback helps the model become smarter over time.
The solution’s intuitive toolset lets subject matter experts without coding or deep learning expertise manage the models and train new ones. This reduces clients’ personnel costs and encourages them to explore new applications in inspection, worker safety, maintenance, video editing, sample analysis, and other areas.
At our manufacturing facilities that use Visual Inspection, we’ve hosted clients from various industries to demonstrate how we’ve gained a deeper understanding from terabytes of data contained in images and video. The result is smarter manufacturing that helps fulfill the promise of Industry 4.0.
AI gave birth to ML through which a large amount of unstructured data (Big Data) is interpreted. With such data interpretation, ML gives useful information from that data by eliminating futile elements.
Can your file transfer infrastructure handle the implications of different data privacy regulations? Is your network and the data that travels across it secure enough?
Explore how applying AI and automation to IT operations can help SREs ensure resiliency and robustness of enterprise applications and free valuable time and talent to support innovation.