As we begin to feel the impact of the fourth industrial revolution, manufacturers have to look carefully at new, disruptive technologies like artificial intelligence (AI), data analytics, or Distributed Ledger Technologies (DLT) like blockchain.
The majority understand that their future depends on these smart technologies. For some, they will act as a driver for their business; for others, they will be an existential threat. Doing nothing about them is not an option.
But manufacturers are experts at making things. They’re not typically experts at IT. Often they don’t have the technological savvy they need to make Industry 4.0 work for them. At the same time, lots of tech vendors don’t have a strong enough understanding of the manufacturing vertical to properly guide their customers through the digitalisation process as it pertains to this sector.
In addition it remains unclear very often what the exact business impact of a particular technology will be.
Herein lies the challenge: what’s the best way for manufacturers to bring new technologies onto their shop floor? I think an iterative, agile approach is the answer – and in this blog post, I’m going to explain why.
What does agility mean for the manufacturing industry?
An agile methodology is a step-by-step approach that minimises the risks of unknown technologies.
You start with a feasibility test, and then move through proof of value, integration, production, refinement, finally ending in review. Breaking the journey down into smaller pieces is crucial as it allows you to re-assess the project’s value and purpose at each stage – and make changes, or even stop or try something else. This enables you to manage costs and obtain results even in the midst of complexity.
A strict and inflexible business plan isn’t suited to this situation because it doesn’t account for the amount of change inherent in the process. AI doesn’t just slot into the shop floor: it has to be tailored to the operational technology that already exists, and the processes that make the business run. Working out how to do this takes time, and often you find things you weren’t expecting that require you to adjust your goals.
This is why a methodology based on iterative, agile steps is the best option. Every step involves only reasonably small costs and after every step you can align the project with the business goal (we should better refer to this as a guiding principle), or even adjust it if necessary. This is why the approach is ‘agile’ – it allows for things to change.
In the end, it will lead you to the right goal – even if there was no clear goal (basically only a guiding principle) in the first place.
Agility in action
A really good example of the agile methodology is the digital annealer.
This new technology is important for manufacturers because it solves combinatorial problems – in other words, challenges that involve lots of different parameters that cannot be optimised by simply solving an equation (neither exactly nor by approximation).
Imagine the typical challenge of job shop scheduling: Your task is to find the optimum schedule to run X production jobs on Y machines with different machine sequences and duration times per machine for every job. There is no ordinary solution to this kind of problem. You can’t just plug it into an equation; rather, you have to try out all of the potential combinations to see what works best. But this not an option either because testing all possible combinations with just 26 parameters will take as long as the age of the universe (assuming your computer can test 10 billion combinations per second). The digital annealer computational architecture bridges the gap to the quantum world and hence is able to find solutions to these hugely complex problems in a shorter span of time.
Manufacturers face these kind of complex problems all the time. If you use robots to weld a product, you’ll need to work out the best way to set up the production line so that each robot can do its welding while moving as little as possible from the end of one welding point to the start of the next.
The digital annealer can help you make a start on this problem. But only if you use it in an iterative way.
For the welding issue, the best process would be to start analysing one robot and then decide on your next steps. If you’re happy with how this has gone, you are almost done but it may happen that the results are not yet convincing enough. You would then revisit the problem, and extend the problem scope – maybe covering more robots and developing a cooperation model for multiple robots. This may deliver better results in terms of expected speedups. It is important to understand that you most likely did not have the necessary insights into which way a cooperation model of multiple robots carries a much higher optimisation potential than just applying it to one.
After finding the solution with the right optimisation potential then, you can move into a piloting phase, where you integrate it for the first time in a real environment. After the piloting phase, if you are achieving the results you wanted, you can move into mass application.
All in all, the process can look like this:
- Putting a first problem approach in a mathematical framework
- Run first samples using the DA cloud service
- Verify the impact
- Revisit the problem if necessary
- Adapt the problem approach accordingly
- Verify the impact
- Find a first pilot for real world integration
- Go into mass application
It’s not a strict business plan, but a step by step approach under a guiding principle. Breaking it down into stages allows you to investigate and learn more before you fully commit. Each step involves only reasonably small costs, and gives you an opportunity to align with the business goal.
The second half of an agile approach: partnership
An agile approach can make the digitalisation process far easier. But it still doesn’t solve the big problem for most manufacturers: a lack of technological savvy.
This is where it becomes crucial to undertake this process in a partnership. Most manufacturers will need a partner to help them with the finer points of the technology.
A co-creation approach brings all the relevant experts inside and outside the company around one table. So the organisation looking to adopt new technologies can learn from a partner with a wealth of experience in technology and in digitalisation projects more generally.
A big part of being agile is learning to take on the advice of others. That’s why the final part of an agile methodology has to include partnership and openness to collaboration. Without these principles, even the most technically aware manufacturers will struggle.
Seizing the opportunity in Industry 4.0
Disruptive technologies are on the horizon. Manufacturers can’t change this. Instead, they have to seize the opportunity that’s being presented to them, and prepare for it.
This means bringing technology into their organisation in a way that works for them, and the end user they are trying to serve.
An agile methodology is the ideal way to go about this. It allows a manufacturer to make small iterative changes, so they never lose sight of what they want out of their new technology. They can control costs while making big, serious changes. And it can all be facilitated by the right partner.