Profile: Transforming Connectivity with Sensor Technology

In a word, I would say sensors are ubiquitous. They're ubiquitous and improving our lives across our work, our play. Examples that I think about that people are familiar with, they're following their sleep through using an Oura ring, for example, or maybe tracking their heart rhythms on their Apple Watch or they're getting driver assistance in new systems, installed in the latest cars. We’re watching traffic flow so that traffic can move a little bit more smoothly through the cities. Room lighting is being controlled by various sensors so that we are minimizing energy consumption. There are a lot of different ways our lives are being impacted by sensors. Some we know about and some maybe we're not yet aware of.

There are a few trends affecting where sensors are being placed and how they're being used. The first one I think about is sustainability. Today we see a lot of electric vehicle adoption. We have sensors involved in battery charging systems, so people don't have to be standing in parking for too many hours to charge up their car when they're on a road trip. I mentioned before about the room lighting or room heating to try to conserve some energy usage, in office buildings, for example. You also use sensors for a lot of feedback. In industrial environments you have artificial intelligence or machine learning being used to monitor important industrial assets like pumps or motors for preventive maintenance reasons. And then another trend that's shaping the use of sensors is remoteness. Right through this Covid period that we've all been living through, we've learned to work remotely, but other things are being done remotely as well, even something as complex as remote surgery. We have doctors conducting surgery remotely. You need very fine sensors to provide feedback to the surgeon that's far away, perhaps from the patient. Sensors are enabling a lot of these types of new use cases that are following up on the trends that we just talked about.

This is an area where we're focused a lot these days in TE Sensors. I think many people, for example, are familiar with pulse oximetry or the blood oxygen levels that are being measured, particularly for patients in the hospital with Covid right now. And that turns out to be an optical sensor of some type. It's a very basic sensor. We continue to develop beyond pulse oximetry. We do sell sensors for that but we're also doing things like developing piezoelectric films and devices. And these piezoelectric films can act as a dynamic strain gauge. How does that work? Well, that can monitor your heart rate and your respiration when you're sitting in a chair or lying down on a hospital bed, so you don't even have to have a blood pressure cuff or somebody there counting your breaths. You can have your body motion detected from breathing or from your heart rate pushing against your skin. The piezo sensor is extremely sensitive to motion, which enables this capability.

Another use of such a piezo detector is for pets. There is a trend toward increasing medical care for on our pets these days. A lot of times we have to take them to the doctor and they're a little bit harder to get a pulse or a respiration rate to have them hold still. But if you place your pet on one of these mats, you can now detect their heart rate and their respiration rate using piezo monitors. In addition, in a different kind of sensor that we're making here at TE Sensors are very small, miniaturized pressure sensors. And these are used in micro surgeries, things like kidney procedures where the pressure within, kidney blood vessels is really critical for successful surgery, as well as in ocular surgery. When people have cataract surgery, for example, you could think about that there's some pressure involved in replacing the lens of a cataract and some other ocular surgery. There are very important applications that require extremely precise pressure measurements in the body for successful surgery. And TE is on the cutting edge of some of those applications today.

In the in the industrial space, we talked about feedback being a theme for using sensor data to make an improvement in whatever process you're looking at. In the industrial environment and in cobots, we're trying to automate something or have a robot working alongside a human to accomplish a task. And automation in general will improve the repeatability and improve the quality of any task that's mass produced. That's the goal, how do we make it even more repeatable and more precise. The cobots that we're working on, we can create a force sensor or a torque sensor that can be placed in a cobot arm and will be very precise in terms of when the arm is articulated or touching its target. This level of safety that we're providing out of the cobot is beyond the current sensing capabilities of today's cobots. We are improving the ability of a cobot to interact with the human, making the human safer, making the cobot much more precise with its actions. That's certainly one application.

In general, in industrial automation, you want to maximize the uptime of your industrial assets. Predictive maintenance becomes a prime way to do that, rather than having a maintenance schedule where every two weeks you take the asset offline to do some preventive maintenance or lubrication or whatever it is that you're going to do to maintain the asset. Now, you use sensors to tell when the asset needs maintenance and you develop using, sometimes you use machine learning or AI techniques in order to analyze the data that you're getting from the asset. But once you have that data, you can much more finely tune your predictive maintenance schedules and keep your asset working for you to provide more uptime.

I tend to think of vehicle autonomy as kind of the holy grail of sensing. There are so many sensors that are that are required in order to enable autonomous driving. Part of the challenge is that you've got to not only detect or sense what's going on in the entire environment around you as you're driving, but you've also got to be able to distinguish what those objects are given the particular environment that you're in. It's very situationally based, as well as very complex in terms of the kinds of changing environment. I'll give you an example and I hope everybody can understand this, but it's a little bit of an older one. We used to have a pretty popular computer brand called Gateway Computer, and their brand was signified, their boxes and their trucks came looking like Holstein cows. You had these black and white sort of patterns on their boxes and on the sides of their trucks, and they really looked like a cow, a dairy cow. You could imagine now if you were driving alongside a gateway computer truck that looked like a cow. And you were being guided by a lot of sensors. What would these sensors interpret this truck to be? Would it say, oh, you're driving next to a cow, or oh, you're driving next to a Gateway truck? The amount of training and the amount of compute and machine learning that's required to take the sensors and interpret the data in the environment that you're in is just staggering.

And this to me is really the challenge with full autonomy here is not only is it sensing by itself, but it's that interpretation or distinguishing in the given environment. Now you have to do it at speed. You have to solve that problem and you have to do it in very few milliseconds in order to enable the right response time, whether it’s your braking response time with your foot or the machine response time that you have to build in. So those are all the challenges that need to be overcome to provide full autonomy in vehicles. Meanwhile, I think you can improve the driving experience with sensors. You've got your in-cabin temperature and some of your awake sensors to make sure your driver awake level is being monitored. You've got some external headlight beam directional sensors that can help you automatically make sure your headlights are aimed in the most useful direction for while you're driving. Those are a couple of examples that are already in play today. You can see the gap between the things we're doing today, which are valuable. But full autonomy is still some ways away. 

You just named off a whole bunch of applications that we are all excited about. We do have a lot of great target markets here. I would say from the TE side, the way that we are approaching these is first, what we do is we improve the precision and the accuracy of the base sense elements. We have a very broad array of products that basically take some measurable element in the environment and turn it into an analog signal, whether that's a pressure signal, a temperature, a position monitor. It could be vibration, it could be an optical or photonic signal. First, we have a whole array of sensing elements that are world class. We have that to begin to base a lot of products on. Next, what we want to do is combine some of those sensors to create even more valuable outputs for our customers. An example there would be to combine a temperature sensor with a pressure sensor in order to create a calibrated output. Most of our customers, when they buy our sensor, they need to have it calibrated to a particular scale or particular precision level so that they understand the range or the dynamic range in which they're operating. Multiple sensors helps to provide an on-board calibration capability that adds value to our customers. That's the second step.

Then the third step where we're headed with TE sensors is to now work in a larger partner ecosystem to fully embody the industrial internet of things. We want to combine our sensing outputs, the calibration that we've talked about, some type of communication, whether it's wired or wireless, and ultimately also machine learning and AI to create models. Earlier I talked a little bit about industrial asset performance, models of asset performance or other performance models that will solve problems for customers. Whether those are industrial pumps or pipelines, transformers. You talked about many of the different applications, but what it’s basically taking a sensor output, creating a digital form of it, perhaps combining it with other sensing elements to create an even, a more complex but more valuable output, and then combining it with communication and machine learning to take those outputs and move them to where the customer can receive them and act upon them. This is the direction we're headed in TE Sensors. We’re all pretty excited about it. I think it's a great opportunity, and I think you're going to hear a lot more from TE in this space in the coming months and years