Last week I had the opportunity to give a presentation about IoT and condition-based maintenance at the 2018 Technology & Maintenance Council (TMC) Annual Meeting & Transportation Technology Exhibition, hosted by the American Trucking Association in Atlanta, GA. TMC represents some of the best minds in transportation and drives consistent improvement throughout the trucking industry.

My presentation focused on optimizing condition-based maintenance with IoT technology, as well as a look at best practices and potential standards for the trucking industry. Why wasn’t predictive failure – one of the great promises of IoT – a greater point of emphasis? The answer is that, much like those jet packs we were promised as children, there is a difference between what is possible and what is pragmatic.

That’s not to say predictive capabilities aren’t possible. I work with data scientists every day and we do, in fact, combine their skills with technologies like machine learning to predict all sorts of things with confidence. The problem in the industrial space is that, as many thought leaders have recently noted, the data often sucks or the business case isn’t clear.

In fact, I’ve seen many organizations turn what is essentially data-mining into a roadblock for all sorts of other valuable objectives that are readily achievable. This is why I suggest the more practical approach of taking actions in response to combinations of current conditions now as a prelude to greater IoT success – such as predicting vehicle failures – down the road. To accomplish this, I recommended the following:

  1. Forget about technology (but only for a moment). As a software company, this may seem counterintuitive. But deploying IoT without a clear vision of the end goals or how to measure success is a recipe for disaster.
  2. Pick a business problem and objective. A clear definition – such as reducing maintenance costs by 5% by optimizing work during a scheduled downtime – is essential. An aimless initiative without a set methodology for measuring value doesn’t help anyone.
  3. Identify current diagnostic IP and subject matter experts. People with the training and knowledge to not only identify and diagnose problems, but also prescribe a solution, are critical. This can include engineers, mechanics, and maintenance teams, as well as operators (i.e. drivers) and customer support. Getting the right people involved at the right time, in the right capacity can accelerate almost any IoT exercise by operationalizing what your team already knows.
  4. Be brutally honest. Are technological capabilities mature enough for IoT? The only wrong answer is an inaccurate one. Have a clear understanding of the underlying capabilities necessary vs. actual capabilities (e.g. how data gets from truck to cloud for analysis and real-time monitoring) and what’s needed to scale before getting more ambitious. Recommendations 1-4 should drive this process, but be careful not to box yourself into a technology that won’t support future use cases. Essential in this conversation is an assessment of current data quality as it pertains to the use case and a willingness to put analysis of historical data on a separate but closely related track, rather than as a milestone. Analytic insights can turn into rules for specific data events at a later time.
  5. Get started. The right partners and guidance are important, but recognizing and seizing opportunities begins with organizational motion and momentum.

A practical approach can help minimize organizational and individual resistance, as well as build the foundation for more advanced capabilities like artificial intelligence, automation, and on-board intelligence. All of which can greatly enhance the potential for IoT success. Based on feedback I received at the conference, there was a sense of (almost) relief that a roadmap exists for traditional industrial organizations to reach new goals and remain highly competitive with IoT.