7 pieces of advice from automation experts on AI, edge computing
Control Engineering subscribers answering survey questions on “Edge computing and artificial intelligence technology” research also provided advice on how to use edge computing and AI technologies.
- Understand Control Engineering research advice about use of edge controllers, artificial intelligence and machine learning (AI/ML) for automation, controls and instrumentation.
- Compare advice on AI and edge computing to your knowledge and experiences.
Edge computing/AI research insights
- Control Engineering research on edge computing and artificial intelligence (AI) provides advice from survey respondents about how those in automation are applying each.
- Additional advice on edge computing and cybersecurity concerns about vulnerabilities also were highlighted by automation experts.
Advice is available from the “Edge computing and artificial intelligence technology” research. Control Engineering subscribers took the survey February and March 2023 and provided advice about how edge computing and AI are being used for automation, controls and instrumentation. Survey respondents were asked to “Please provide explanation/advice about using/applying edge computing and/or AI.” Among respondents, 41 wrote in related advice, some of which follows, lightly edited for style. Appearing first, the seven bits of advice providing the most detail on AI and edge computing.
Advice on edge computing for automation, controls, instrumentation
See seven pieces of advice on AI and edge computing, followed by additional advice.
Think of these tools as complementary to basic process control systems – not as alternatives to control systems.
We are using edges computing to connect to programmable logic controllers (PLCs) and send us back data about the performance and health of fans and compressors.
AI and machine learning (ML) can be used to predict maintenance needs, improve overall manufacturing efficiency, optimize production and predict demand.
We are finding it better to write our own code and develop our own models for both vision and controls.
Edge computing device can handle secure communication outside the fence while providing specific information inside the fence.
We have integrated powerful edge-computing capabilities, open software and hardware resources, and software development kits (SDKs) with strong engineering field support and rapid application deployment support.
Really consider how and why you are deploying computing power out at the “edge” in industrial environments. It is not always the correct solution for the problem.
Other AI, edge computing advice from Control Engineering research
We use edge computing for certain software applications for easier deployment and software access for multiple technicians, as well as using edge devices for data transactions and collecting information.
Edge computing is old-fashioned modular computing in a world of easily accessed external modules via advanced applied programming interfaces (APIs). Stop making it more than it is.
AI complements mathematical optimization models in operations planning.
Caution about difficulties using edge computing devices with control systems due to poor cybersecurity of edge computing devices for all the vendors analyzed. The edge computing devices are still very much information technology (IT) and have not been designed for secure integration with industrial control systems.
Leverage hardware vendors for their specific hardware.
Automated decision making and security are the most import reasons.
Use AI to improve machine vision for visual recognition in a manufacturing environment.
These can be overwhelming to conceive, but become easier as you apply them.
We are understanding the capabilities and necessities or our systems.
ML can improve autonomous vehicles.
AI is the wave of the future. It is wise to get on board this train.
Methods for the 2023 Control Engineering edge computing, AI research
To do the research, subscribers were sent an email from Control Engineering asking them to participate in this study. The email included a URL linked to the questionnaire. Information was collected from Feb. 8 through through March 23, 2023. With 121 respondents, the margin of error for the results was +/-8.9% at a 95% confidence level.
Mark T. Hoske is Control Engineering content manager, firstname.lastname@example.org. Amanda Pelliccione, director of research and awards programs for CFE Media and Technology, conducted the research and assembled the related report, including much of the information above from the executive summary.
KEYWORDS: edge controller advice, AI/ML advice for automation,
How are you applying edge computing and AI/ML?