Sensors veteran Roger Grace spoke with EE World about his award, data acquisition, IoT, and the trouble with AI.
Early in my editorial career, I met Roger Grace and knew then that he was someone special in the sensors community. About 30 years later on June 25, 2025, Grace received a lifetime achievement award at the Sensors Converge conference for his contributions to the sensing industry. EE World spoke with Grace on August 14 about his award and how the sensing business has grown and changed in importance, especially in the age of AI.
In the dim and distant past, we used data-acquisition systems, then IoT, then medical, and now AI consumes every bit of data we can produce. Without physical sensors, we can’t measure our world no matter how much AI takes over.
Numerous sensors connected to data-acquisition systems that conditioned signals, digitized them, collected the results, and stored the data. Computers using software from spreadsheets to those with data analysis and plotting features then display the data in tables or plots.
Those sensors included thermocouples, strain gages, pressure sensors, and flow sensors. Grace cited a typical example when he consulted for Nova Sensor early in his career. In one instance, engineers at Pressure Systems Inc. (acquired by TE Connectivity) were trying to measure forces on an airfoil while it was in a wind tunnel.
“Because you had this airfoil in a big wind tunnel, you had to make lots of measurements and quickly make them. This company used a manifold called a “scanner,” which is a piece of metal that has 36 inputs. Each input had a pressure sensor. The manifold had plastic tubes embedded into or pasted onto the airfoil. That pressure pulse would go down the tube into the pressure sensor chip, which sent an analog signal over a wire to a data acquisition system that did all the signal conditioning and analog-to-digital conversion in the box. They were using masking tape on the air foil to tie down the wires that carried the electrical signals over the wing into the data acquisition system.”
From DAQ systems to IoT
Sensor-based measurement systems have evolved as microcontrollers and ASICs combined decision-making and control capabilities with digitizers, which brought the electronics closer to the measurement point. Instead of wires carrying analog signals, measurement systems began sending digital representations of the measurement to a host computer, initially over buses such as RS-232/RS-485.
As sensing systems added wireless communications — Wi-Fi, Bluetooth, Cellular, LoRaWAN, and others — they evolved into IoT devices, something Grace claims has not taken off the way people thought it would. That is, IoT hasn’t kept up with the hype it generated, though he admitted that he needs to do more research into that.
Grace called the introduction of the Analog Devices ADXL50 an inflection point in data acquisition because it integrated an accelerometer with signal conditioning into a single device.
Sensing has also penetrated into many fields outside of industrial measurements. Grace noted that medical sensing has greatly expanded, not just in hospital equipment but into personal sensing as well. In particular, Grace cites wearables that include those you attach to yourself.
“A wearable is a data collection platform that happens to be on your body,” said Grace, “whether it be a T-shirt, a pair of socks, a hat, or a watch. It’s a data-collection platform that’s part of IoT. If you don’t have a sensor, you have nothing.”
That includes AI.
“Wearables are, in my opinion, the basis of IoT and AI. When you look at the opportunity that wearables present, especially from a sensor point of view, a wearable doesn’t have any intelligence unless it has a sensor.”
Grace noted that “AI is sucking the oxygen out of everything.” By that, Grace explained how investors in technology companies want to see an AI component. “Everybody who wants money has to say what they’re doing with AI. If they don’t, nobody wants to give them any money; they don’t think there’s any future in technology unless it’s AI-based or AI-driven.”
AI needs data to analyze. As Grace concluded, “The more sensors that you get inputs from, the more accurate you can do post-sensing versus pre-sensing. You start taking all the data, and you look at trend lines. Then, you start doing correlation coefficients to get highly accurate data. That’s what, in my opinion, is going to drive AI from a sensor point of view.”
Filed Under: AI Engineering Collective, AI • machine learning, Sensor Tips, Medical-device manufacture