Constant Change: An Industrial Engineering and Manufacturing Perspective

Change is the only constant, as the saying goes. Change Notice, a weekly video podcast hosted by Instrumental Inc., seeks to address the topic in conversation with product design engineering leaders. In a recent episode, Maya HTT’s Remi Duquette joined host Anna-Katrina Shedletsky for a deep dive into the changes he’s seen over the last 10+ years. This is a summary of their conversation:

Accelerating and layering complexity

From Remi’s perspective in the industrial engineering and manufacturing space, the pace of production that has accelerated in ways that would have been inconceivable even five years ago.

The pace of change, the complexity of systems and software, the advent of industrial AI, and the unprecedented need for collaboration are all contributing to the changing industrial landscape.

AI has introduced a new layer of complexity as AI models and self-learning run on top of electronics, on the edge, as part of the product. Traceability is becoming very complex. As a result, development teams and manufacturers must keep on top of products that continue to evolve long after the product is out the door.

Industry 4.0, data, AI: Broken promises and snake oil?

Manufacturing has changed, operationally, over the past decade in significant ways. Early players in data and AI promised the moon and then some, fueling unrealistic expectations, swiftly followed by repeated failures, disappointment and disillusionment.

Maya HTT takes a grounded, small-wins approach guided by business use cases. Many companies understand the value of their data, but most still struggle with how best to use it and how to operationalize AI for their business. Unrealistic big-vision expectations don’t help.

[bctt tweet=”Industrial #AI requires a grounded, small-wins approach guided by business use cases” username=”MayaSimulation”]

Guiding principles of data

•  Machines evolve faster than people: Data collection has progressed at lightning speed; collaboration has not evolved at the same pace. Silos remain a challenge in many organizations.

•  Don’t think too big. Start small to rapidly add value and see success, then set your sights on scaling up.

•  Data is dirty until proven clean. With this mindset, you will be well positioned to build the right process and engineering pipeline for reliably successful outputs.

•  Dirty data lakes don’t yield a lot of value. Focus forward on collecting clean data rather than on retroactively cleaning dirty data. Identify the small chunk of data you need to clean; find out where it got dirty and build safeguards to ensure it comes in clean in the future.

•  Take the shortest path to leveraging data for ROI. Use what you have in place; don’t start by investing in a rebuild for the first use case.

[bctt tweet=”Data is always dirty until proven clean #data #AI” username=”MayaSimulation”]

Partnering for AI

When looking for an AI partner, don’t settle for a jack of all trades. Industrial data applications are a specific expertise. Your data isn’t generic – a physics-based understanding is needed in industrial manufacturing. Generalized solutions typically fall short.

It’s worth investing the time to drill down to a provider’s practical accomplishments. Look for previous experience solving the exact same or highly similar problem as yours, and request real examples.

A dose of skepticism is healthy. Ask questions. For example, what does it actually take to get up and running? Roughly how much time would the proposed solution take?

Maya HTT’s solutions for the changing landscape

1. Creating simulation models: AI-based reduced order models and virtual sensors make it possible, for example, to place virtual sensors where it would be physically impossible to place real sensors.

2. Building actual AI reinforcement models to change controls: For example, of an electronics system. A reinforcement agent will learn from the simulation and virtual environment and carry that learning into the real environment. In a sense, it gets a head start.

3. Building AI software as a service (SaaS): Maya HTT offers services to augment clients’ manufacturing or operations, as needed.

Maya HTT has the industrial physics-based expertise and experience to guide you to the right solutions and approach for getting the most out of your data and seeing quick wins you can build on and scale up.

Subscribe to our newsletter