IOT is old, meet AIOT
Next month I have been asked to present at the Industrial IOT Summit, the event is sponsored by the giants of industry, GE, Schneider, Bosch etc. As I sit here writing my presentation, I'm reflecting on the phenomena that is the Industrial IOT. Firstly, why the fuss? Sensors have been around for decades in industry. Many may be wondering what all the fuss is about. Isn't IOT really about RTU's, PLC's, IMU's (and other 3 letter acronyms?). Well yes,.... and no.... Let me explain.
We've seen huge advancements in consumer technologies over the last 15 years, driven by the internet, mobile phones, telecoms, cloud etc. The underlying technologies that drove the consumer era are now creating huge opportunities across industry. We can now apply extremely powerful (but low cost) sensors, transmit massive data, store it in the cloud and analyse it with Artificial Intelligence (AI). The opportunity now is to have machines analysing machines.
Sensing conditions such as temperature, vibration, humidity etc is really just measuring, it is a value (at a point in time). When analysed by a human, context is added. Measure + context = meaning. For example, if a machine temperature is 50 degrees, is this good or bad? Well, it depends on the context, without context, it's impossible to answer, we need to know what's normal, from there we need to understand is this figure rising or falling, if so, how quickly? etc. (You get the idea.) At present it seems most IOT vendors are so focused on the measures, they miss the meaning. Decisions are made on meaning, not measures.
Unless you've have been hiding under a rock, you will note the massive predictions of billions of sensors being connected to the internet. The general consensus is that for every human (we now have 7B), there will be around 7 IOT devices (or ~50B). So who exactly is going to provide all of this meaning ? Well, unless we start to employ every man, woman and child ...
So the question remains ? At MOVUS, we've thought a great deal about this and as a result we've developed an IOT Value Model. We see four key stages of value creation. Firstly Visibility (where is my asset?), Utilisation (is it operating or not?), Availability (how healthy is it?) and finally Life Cycle Cost (purchase price+maintenance cost+energy+disposal). Visibility and utilisation are fairly straight forward. However, these are the low hanging fruit, the real value is in availability and life cycle cost. No surprises, energy is typically 50-60% and maintenance is 25-40% of the costs of a machine. This is where it's gets interesting. Imagine when purchasing your new pump, if you knew the meantime to failure for that pump (measured not estimate), or you knew the expected failure modes or you could compare the total life cycle costs (purchase price, average energy costs, average maintenance costs and disposal costs) between makes and models? Would purchase price be so important anymore?
Maintenance practices have largely remained the same for decades. Simplified, run until it breaks (then repair/replace), inspect periodically (repair/replace), replace ahead of expected failure and for a small amount of cases (<5%), put a expensive array of sensors to capture every detail about the machine. The first three cases require humans, the last, uses lots of instrumentation. So where does IOT come in?
If we are to achieve the value of the IOT revolution, then mankind cannot be the bottleneck, in providing meaning to measure. In achieving the real power and value of the IOT, we need artificial intelligence systems that analyse, predict and ideally learn. At MOVUS we've building such a device (the FitMachine - yes we have the trademark - sorry Fitbit, burn!). With our device installed, technicians don't climb on rooftops or down pipes filled with sewerage (nasty!) just to inspect machinery.
The real value of IOT isn't in the measuring, it is in the ability to make more informed decisions. For value to be achieved then decisions need to be made. To reduce the total cost of your assets then I believe AI is the only way forward if we are to adopt these technologies on a global scale. The benefits that flow will be improved safety, reduced risk, reduced cost and our machines will be operating longer. Oh!, and the most important benefit, people who feel empowered with better information to make far more informed decisions (which they didn't have to climb down a sewer pipe to get). :)
NOTE: If you are at the summit, come and say hi, mention this article, we'll give you a FREE trial of our FitMachine for a couple of months.
Brad Parsons is CEO and Founder of MOVUS, are building the FitMachine - 'A Fitbit for Industrial Machines'. Find more information at www.movus.com.au
*FitMachine is a trademark of MOVUS ** Fitbit is a trademark of Fitbit.com
Helping clients with the strategic partnerships that Integral has to deliver enterprise solutions to our customers. Trust, Collaborate, Co-Create, Delight.
8 年Great article Brad. Once connected and analysed, we then have the opportunity to optimise
Account Manager
8 年Great points Brad. The question is, what platforms are out there to securely link all the IoT devices so we can gain the benefits of the AI?
Senior Solutions Architect at BlueScope Steel Australia
8 年Am working in a parallel AI field at present, in application of neural marketing where marketing strategies are developed through machine learning based on behavioural analysis. On a semi-regular basis I get gentle nudges from my right brain suggesting we are behaving like creative lemmings with respect to AI. Anyone else get these disquieting nudges? I note that John Hanna's comment is somewhere along this vein.
Executive Leadership ? Chief Information Officer ? Co-Founder Non-Executive Director ? Passion for Transforming Complex Ecosystems ? Purpose and People ? Aficionado of Horological Artistry and Alfa Romeos
8 年Excellent insights. As there is IOT & AIOT, one also of importance is VOE - Vulnerability of Everything. As we get smarter & more connected, the exposure becomes greater unfortunately.
Partner/Owner at Indago Partners
8 年Brad, Good luck with the conference talk - looks like you're one of the few true Startups in the speaker lineup, so what you have to say should go over well. I'm not so sure about the continual creation of new acronyms to describe what's the "latest thing". I do agree, however, that adding context to data and using advanced analytical techniques (especially AI) are definitely where the value lies in all this and that it's not just about more sensors in the field and more data managed in the backend. This is not really a new thing, but has been building for some time. In the Resources industries, I've seen this happen a couple of times before. In the early 1990s, minerals exploration was being revolutionised by the combination of more data (eg. cheaper geochemistry over a wider range of elements, new remote sensing geophysical instruments) and better tools for managing and analysing the data (eg. relational databases, GIS, Image processing systems). In the late 1990s the petroleum industry was going through their "digital oilfields" revolution, also with new sensors bringing much more data (eg. measure-while-drilling instruments, 4D seismic) and new methods of analysis (e.g. 3D Cave visualisation, geostatistical modelling, remote operations). I do think the latest revolution is different, but mainly by the scale and complexity - the shear number of new sensors in the field AND the number of new methods of analysing the data AND the number of different business processes and industries being impacted at once. By the way, it's worth having a read (or re-read) of the seminal report that Ed Luczak wrote in 2005 for the CSC LEF - https://assetsdev1.csc.com/it/downloads/13909_3.pdf. He outlined three major data trends that were leading to unprecedented opportunities: - Data Everywhere - data in many places, changing the rules - Time and Place - data about when and where people and things are, and what's happening now - Social Connections - data that strengthens connections between people - Meaning - data that helps make sense of it all (metadata, semantic web, data standards) All these trends have continued unabated since 2005, so perhaps we're just now reaping the rewards across a wider audience. I seems to me that the last trend (Meaning) has been lagging the others and is likely to become a major bottleneck to progress. For example, in the equipment reliability space, if everyone is going to describe things differently (e.g. different time-usage models and different down-time/event categories), then it makes it hard to develop reusable solutions and hard to leverage the advances from one area to another (one site to another, one system to another, one industry to another, etc). That is, we need to get more serious about describing the CONTEXT in consistent ways. Also, it's not so much that maintenance practices have largely remained the same in decades, it's that we've seen incredibly different levels of maturity across different industries. Certainly some industries have been stuck on a low level of maturity, which is perhaps a different story. You only have to compare the application of condition monitoring in different industries. Maybe now that we have better and cheaper sensors and better and cheaper AI then this will help drive cross-industry adoption faster, but I expect that the lack of standards in the description of Context will be our biggest issue.