Gathered data turns into valuable information when systems and people can quickly make sense of it and put it to good use – which is where artificial intelligence (AI) steps in.
Mainstream AI applications, such as automated speech and image recognition, solve problems and make sense of the world by looking for common patterns in easily accessible and interpretable sets of big data. Here, when we talk about big data sets generated by mass-market consumer applications, we mean millions or billions of entries.
But industrial AI typically has to deal with much smaller data sets and this presents a real challenge when it comes to identifying norms, patterns and anomalies. A common task for manufacturers is to detect, diagnose and predict rare events with negative outcomes so they can effectively avoid them in future. Or when processes go extraordinarily smoothly, manufacturers will want to understand how they achieved perfection so they can optimise processes in order to repeatedly replicate their success.
Interconnected devices and machines in industrial environments produce sizeable quantities of sensor data. For AI systems to efficiently find patterns in the data it is essential that the data is organized properly with a logical storage structure. Sensor data also tells a much fuller story for diagnostic and predictive purposes when it is enriched and fused with additional data sources, which can be drawn from design, manufacturing, operation and servicing processes, the surrounding environment, or neighbouring systems.
In addition, because industrial applications are complex, it is crucial to factor in human interaction when creating autonomous systems. Not only should a human be able to override an AI-based finding for safety purposes – think of steering a self-driving car – but human input is an important source of feedback that helps to retrain and hence to improve AI models over time.
On 13 February 2020 I’ll be speaking at the Smart Maintenance conference in Switzerland about how advanced, continuous and remote analysis of sensor data can provide a viable and cost-effective way to automate and enhance multiple applications across a variety of business areas, including the manufacturing, construction and mining industries.
Hexagon is well placed to develop and support industrial AI solutions that are adapted to the specific needs and constraints of manufacturers: Since Hexagon is one of the few companies that operate in both Operational Technologies (OT) and Information Technologies (IT), we are equally at home deploying engineering- and data-driven approaches when creating autonomous connected ecosystems that leverage AI.
Our data-driven products and services range from reality capture solutions, positioning technologies and location intelligence, through to advanced design, simulation and production software and a depth of metrology solutions. As a result, we have been able to draw on our expertise to develop AI solutions that include condition-monitoring and predictive maintenance of metrology equipment, as well as industrial assets, vehicles in mining, and critical infrastructure for utilities. And it is a journey that is far from over as we continually develop new autonomous services to drive smarter manufacturing.
