Expense calculations in the industrial world tend to be unintuitive to individuals, because service and maintenance often consume larger chunks of the budget than the equipment itself. For that reason, modifications to equipment are usually seen as too risky to justify — they can too easily impact serviceability. So, what do you do when you want to bring an old piece of equipment into the modern age? Michael Bryan Ross’ solution was to use AI to look at analog gauges.
Ross wanted to address a simple and common problem: the equipment has an analog gauge and it would be nice to have that value available in digital form for monitoring and logging.
Most of us, when presented with that problem, would take the easy and seemingly reasonable approach. That might be something like replacing the analog gauge with a microcontroller outfitted with an ADC (analog-to-digital converter).
But very few plant managers or manufacturing engineers are going to give the green light on a modification like that. Not only is there upfront downtime to consider, but it also puts the equipment and future serviceability at risk.
Ross’ solution, on the other hand, is much easier to approve. That’s because it doesn’t require any modification to the equipment at all. In fact, it doesn’t even need to make physical contact with the equipment.
It works by using a camera and AI running on the edge to read the analog gauge. In this case, “the edge” is an inexpensive Arduino® UNO™ Q and it looks at the gauge through a standard USB webcam. The UNO Q runs a MobileNetV3-Small model through ONNX Runtime.
To test that — and to gather the images needed to train the model in the first place — Ross built a physical device with a real analog gauge driven by an actual pressure sensor. To create a training data set, Ross simply collected a bunch of images of the gauge’s needle in different position, then had GPT-5.6 read the black ticks to classify them by numeric value.
Ross acknowledges that the resulting model isn’t perfect. In particular, it tends to lose accuracy at very high and very low ends of the gauge range. But that is a fixable problem (largely through training). The concept holds: that this approach makes it possible digitize analog gauges, without spending much money and without modifying equipment.
Facts Only
* Michael Bryan Ross developed a method to digitize analog gauge readings.
* The system uses a USB webcam and an Arduino UNO Q.
* The hardware runs a MobileNetV3-Small model via ONNX Runtime.
* The process requires no physical modification or contact with the equipment.
* Training data consisted of images of a gauge needle in various positions.
* GPT-5.6 was used to classify the black ticks on the gauge by numeric value.
* The test environment utilized a physical analog gauge driven by a pressure sensor.
* The model exhibits decreased accuracy at the very high and very low ends of the gauge range.
Executive Summary
Industrial maintenance budgets often prioritize serviceability over equipment upgrades, making physical modifications to old machinery risky and difficult to approve. To bypass the need for downtime or hardware alterations, such as replacing gauges with microcontrollers and analog-to-digital converters, a non-invasive edge AI solution has been implemented.
By positioning a USB webcam to view an analog gauge and processing the feed through an Arduino UNO Q running a MobileNetV3-Small model, numeric values can be extracted and logged digitally. The training process leveraged a combination of physical pressure sensors and GPT-5.6 for data classification. While the system currently experiences accuracy drops at the extreme ends of the gauge's range, the approach demonstrates a low-cost method for modernization without compromising equipment integrity.
Full Take
The strongest version of this narrative is that "invisible" upgrades—those that wrap around existing infrastructure rather than cutting into it—are the most viable path for industrial modernization due to the rigid risk-aversion of plant management. It presents a clever marriage of legacy hardware and modern edge computing.
The narrative relies on a specific framing: the "easy and reasonable" approach (ADC replacement) is presented as a high-risk failure point to make the AI solution appear as the only logical path. This creates a binary between "risky hardware modification" and "safe AI observation," ignoring intermediate options like external clip-on sensors or optical encoders.
At its root, this reflects a paradigm of "overlay innovation." Rather than fixing the inefficiency of analog systems, we build digital skins over them. This preserves the status quo of the machinery while satisfying the modern hunger for data. The benefit is speed of deployment and low cost; the cost is a layer of abstraction that introduces new failure modes—such as lighting changes or camera misalignment—which replace the mechanical failure risks of a physical upgrade.
If this were part of an influence campaign, the playbook would involve exaggerating the "terror" of industrial downtime to push a specific hardware ecosystem (Arduino/ONNX). The actual content does not match this; it reads as a genuine engineering experiment.
Patterns detected: none
Bridge Questions:
1. How does the introduction of an optical layer affect the long-term reliability of data compared to a direct electrical signal?
2. In what ways does relying on an LLM (GPT-5.6) for initial data labeling introduce systemic bias into the resulting edge model?
3. What are the security implications of adding networked cameras to legacy industrial environments?