At the recent IMC virtual Summit on Industrial IoT, Mike Hibbett, Solution Architect at Taoglas Next Gen IoT, spoke on how edge computing will revolutionise smart factories. Readers can view recordings of the two day event, which includes 10 hours of programming, keynote address, and workshops. Recordings will be available for six months after the event.

Mr Hibbett is a Solution Architect working in the IoT Solutions division of Taoglas, based in Dublin. Prior to joining Taoglas, Mike was the director of business development at Firmwave. He has more than 30 years of experience developing embedded systems, focusing on firmware, electronics design and device security architecture. He has led product development teams for companies such as Intel, Philips and Thales where he led the UK high grade cryptographic solutions software group. 

Mr Hibbett began by noting that the scope of edge computing can go well beyond a single building or business site.

There are challenges to edge computing, and there are barriers,” Mr Hibbett said. “I see these as quite different.” There are distinct barriers to address, and there are distinct challenges to address, he said.

The introduction of edge computing is not just about dollar amounts on a spreadsheet, there are other advantages too.

Security is one of the least frequently discussed requirements, yet it is one the major barriers to innovation. If it’s not carefully considered, the issues with security failures can be disastrous.

A smart factory should not be seen as just a single site: innovation can occur across an entire organisation, with multiple facilities. We should also consider smart agriculture too, as intensive farming is increasingly big business with single organisations owning hundreds of farms globally. They have similar problems, only it rains on their equipment a lot more.

When one looks at change in an organisation, it is appropriate to look across the entire manufacturing process, and that scope is very wide. Many technical solutions provide improvements to industrial process silos. But when thought through, a technology solution can benefit all these process silos, and the organisation as a whole, providing improvements in operational efficiency in the broadest sense, many small gains amplifying.

We are now looking for many small improvements, and have to look hard for big improvements. But sometimes, big improvements come by surprise, out of the data.

Asking the question — how can technology help — is frequently the wrong approach. It can lead to an undue focus on efficiency improvements. That’s not a bad thing of itself. But some changes may enable a broader range of benefits. These might include: addressing new regulatory compliance; improving workplace safety; or capturing and exploring manufacturing data for new insights.

AI at the edge enables many benefits for the smart factory: it reduces data transfer; with many sensors, data transfer overhead can be significant. Putting AI at the edge can significantly reduce network utilisation, saving communications costs and power consumption.

AI at the edge also reduces latency. Systems can react faster, at the millisecond level. This can ensure responses fast enough to prevent potential damage to equipment and products.

Another important consequence of putting AI at the edge, especially important in Europe, is that it increases privacy and security within a factory site. Sensitive data does not need to leave the sensor. An example: facial recognition data stays within the device.

AI at the edge also creates the ability to sense more than has been previously possible. It can provide deeper insights due to greater access to local and potentially sensitive data, and reduced latency of data access.

Traditional sensors, which typically have addressed temperature, humidity and power levels, do a good job, but tend to do it in isolation. They operate in silos.

By contrast, the next generation of sensors will be connected and far more functional than today’s range of siloed sensors.These have very powerful capabilities, and will play their part in transforming manufacturing.

Among these are robotics-related sensors, which help deliver the right parts to the right machine at the right time as part of JIT manufacturing systems. 

Connected vibration sensors are now appearing that AI systems can utilise for evidence of wear in advance of failure. These allow better planning of preventative maintenance. “Change this while changing that” scenarios.

Vision-based sensors, while already well established in some production processes, cost reduction and power reduction and communication efficiencies move these out of tight silos into the broader operational flow. 

Another is part identification via vision, or RFID. These are addressing manufacturing issues such as: is the tool part where it should be, when it should be? Does this part require traceability through the manufacturing process. These can be extremely low cost passive sensors.

Indoor location services are another developing area: where is this machine tool part, will in be in the right place in time?

Finally, there is the issue of how all the data that can be gathered at the edge can come together to create an ongoing impact on manufacturing processes.

The use of standards-based protocols eases the integration of sensor based data into back-end systems, both in terms of security and in terms of the speed of integration of new sensors.

Smart computing at the edge is critical to reducing the volume of data to be processed. A hierarchy of intelligence within the sensor network will make the design of backend analysis systems simpler and allow the creation of meaningful results much more quickly.

Once data is collected and reviewed, new benefits can arise in very unexpected ways.

A final example: a client of Taoglas who collected IoT data on customer product usage discovered information that transformed their global sales and marketing campaign. This was completely unexpected and unlooked for. But the data on global usage revealed where their main opportunities lay, and had a far-reaching effect on how they targeted their business.