This is one part of a four-part introductory series about autonomous systems. To learn more, read the rest of the series and the free e-book linked at the bottom of this post.
Thinking about adding autonomous systems can seem overwhelming—but our advice is to start small. Identify the places where optimizing process variables can help you meet your operational objectives. When considering your plant’s digital transformation, you may be asking yourself: How could we start with an autonomous system? What use cases make sense for us?
Let’s look at three scenarios where autonomous systems create value.
1. Conditions are highly dynamic
Autonomous systems shine when plants have dynamic process variables and changing conditions.
For example, autonomous systems have successfully helped the proportional integrator controller in bulldozer blades quickly and accurately assess soil conditions (e.g., wet, gravelly, sandy) before digging, and adjust the blade accordingly. This saves the bulldozer operator a trip outside of the cab, where that person would have to physically tune the blade. Not only does the autonomous system help protect the safety of the bulldozer operator by eliminating the trip outside, but it also saves costs and reduces risk for the company. If dirt isn’t completely flat on a construction site, it’s potentially a million-dollar mistake.
2. Humans can’t identify or calculate the best approach fast enough
Humans can only respond to so many variables in a finite amount of time. This means a human has to prioritize optimization that can be accomplished reasonably. However, many processes have variables that even the most skilled experts may not be able to account for—such as humidity or the wear on a machine or part. Humans also struggle in optimization scenarios where the window for optimization is short, but frequent, in duration. Scenarios with multiple variables, many inputs and short but frequent optimization opportunities are all places an autonomous system can help.
Frequently, humans and automated systems work together to control complex processes—particularly in manufacturing. When this is the case, often human operators and supervisors need to quickly assess whether appropriate control is being applied to the process. You can imagine how difficult it is to assess the effect of multivariate control actions on sensor output in real time.
For example, one company Microsoft works with makes medical products. The operations supervisor walks the floor and assesses every five seconds whether the control needs to be adjusted. The only way to perform this assessment in real time is by abstracting to the level of human-readable strategy. Autonomous systems allow the control system to learn and report back human-readable strategy.
3. You want to best use your expert workers
Expert-level workers in your facility have the technical know-how and the institutional knowledge to contribute valuable data and insights to your digital transformation. But experts can’t be everywhere that you need their knowledge – they age into retirement, take too long to train or are scarce in a particular field, for example. Luckily, autonomous systems can capture expert knowledge in a way that’s transferrable and repeatable.
Autonomous systems are built by engineers and trained by your experts who have firsthand knowledge of your systems and processes. They can add their technical knowledge and their on-the-job wisdom—like the sound a machine might make just before going haywire—to help create AI that empowers operators to make better, more consistent decisions without relying on experts as often.
Read more about autonomous systems:
- Autonomous systems 101: Optimizing manufacturing processes with AI
- Why and how to build autonomous systems
- 3 best practices for adding autonomous systems to manufacturing operations
- Free e-book Autonomous Tomorrow: Inside autonomous systems on the factory floor
- Sign up to receive more information about Microsoft Project Bonsai