Verizon Innovation Advisor Tony Lesley of talks about what it means for AI to be “eyes” for Energy, Utilities and Manufacturing.
Tony Lesley:
All right. Hello. My name is Tony Lesley. I am an innovation advisor in Verizon business groups, manufacturing energy and utility practice. I’ve been in the industry for 28 years. The past 19 of those years, I’ve been specifically in our energy and utility manufacturing vertical. So here today to talk a little bit about computer vision. We’ve all heard it said by leadership that we need to put more eyes on that situation, or we need to put a fresh set of eyes on that process, or we need to be looking at this from a different perspective. These are the kinds of things that our computer vision is enabling in massive scale. Think about being able to see with a thousand eyes and a million different perspectives. Processes can change dramatically. Efficiencies can increase dramatically. Quality control can can go through the roof. Safety can be vastly improved.
Tony Lesley:
So computer vision enables a lot of these things that we’re talking about. Let’s talk about those thousands of eyes. What do I mean by that? Quickly you can come to cameras, the thought of utilizing cameras to see different things from different angles, different pieces along a process or manufacturing assembly line. But think about more than just visual. Think about being able to see an infrared. Think about being able to see with LIDAR or radar. Think about other senses. Think about your touch. Think about vibration sensors and adding that data set in. Think about noise. Think about the ability to smell methane gas detection, for example. All of these various sensors can be brought into what we’re calling a computer vision model. So it’s really exciting when you think about the various sensors, devices, things that can be connected and used to build your data learning models around.
Tony Lesley:
The data learning, that’s kind of where I’m talking about the millions of perspectives, looking at things from different perspectives. Are you looking at things from a safety perspective? Are you looking at things from a quality assurance perspective, compliance perspective? All these different perspectives have valid use cases and value. So the ability to build machine learning models for each perspective is amazing. And it’s endless to start thinking about the different perspectives that you can apply to a process or a manufacturing assembly line, or how an end user is using something or how an employee is engaging in a process. Is compliance being followed? Is it a safe environment? Lots and lots of things that are really exciting when you think about applying thousands and thousands of eyes, ears, noses touch with endless number of perspectives. So at Verizon, we feel like there is two core enablers that are going to enable and quite frankly, dramatically accelerate the utilization of computer vision.
Tony Lesley:
Computer vision is real. It’s here, it’s here to stay, but we believe we’re on the cusp of a dramatic acceleration of utilization of computer vision and computer vision models. We feel that there are two key enablers. Those two key enablers are 5G and Edge Compute. So we’re going to talk a little bit about those things today, where Verizon is playing in the space, where the industry as a whole is. And hopefully this will be good information for you to start thinking about how you could utilize computer vision and how fast this type of technology is coming. Is this something that’s so far away that I can’t really try to use it in my environment? We believe, no. We believe that if you’re not already pretty engaged in computer vision, you’re probably behind. So let’s talk about those enablers. We can go to a slide here that talks about 5G.
Tony Lesley:
So 5G, everybody’s hearing 5G, a lot of hype around 5G. It’s amazing the amount of hype, but there’s good reasoning. 5G in my 28 years in telecom is a technology that is advancing things more than any other technology that I’ve seen. This fourth industrial revolution that we’re moving into, 5G is a core enabler of that. When I talk to my clients at refineries or energy generation facilities or manufacturing facilities, the number one problem they have with deploying industry four auto technologies and computer vision specifically is that connectivity. The connectivity is not reliable. The connectivity doesn’t have enough bandwidth. The connectivity is not affordable. The connectivity is not something that I can run my operations on. So I have to hard wire everything in this highly complex, volatile facility atmosphere. It’s very costly to run fiber to all of these sensors and cameras and such.
Tony Lesley:
Was talking just the other day to a manufacturer who has 400 in just one of their facilities. They have 400 cameras today running computer vision. They can’t move those cameras around, they’re hardwired. They can’t deploy them rapidly. They have to take production sites down to deploy, to get the power and wiring to these various cameras. It’s very problematic. So here on this slide, let’s start at 12 o’clock and I’m going to go clockwise around to the different attributes that 5G brings to the table that are different than our current networking technologies of Wi-Fi, fiber or 4G cellular. So starting at the top, let’s look at throughput. If we’re talking about computer vision, you need a lot of bandwidths. You need the ability to really send large amounts of streaming video across the network. You can’t do that with a lot of the older traditional SCADA type networks, wired copper networks that are deployed in a lot of manufacturing facilities. It just can’t support the bandwidth. So that throughput enabler or attribute of 5G is critical for what we’re talking about here today.
Tony Lesley:
Let’s go on around, the next one, latency. When we’re getting to a real time enterprise kind of application where when my camera sees something, I need there to be an action on my assembly line. I might need something to stop. I might need something to adjust on the fly, that requires very low latency. So you have to have a quick agile network to be able to handle that. 5G enables that. 5G along with another enabler that I’m going to talk about, which is Edge Compute and we’ll get more to that later. Reliability, the number one thing I hear from plant managers is this is a mission critical application. I cannot have any downtime. Reliability, 5G brings now five-nine reliability to the network. This is a new step forward with cellular technology, with any wireless technology quite frankly. There’s not another wireless technology that can enable five-nine reliability. So this is a network that you can deploy your mission critical applications on.
Tony Lesley:
Let’s move on around. Service deployment. What does that mean? So that’s the ability to configure each connection for each device, thousands and thousands of devices with the specific attributes they need. Does this sensor need more security? Does it need lower latency? Does it need more throughput? Does it need the ability to utilize less energy because I’m running this sensor on a battery and deployed on a battery? This service deployment enables you to tweak the network for the specifically application or device that you’re working with. It’s an incredible function or attribute that people don’t really think about that often, but it’s incredibly important.
Tony Lesley:
Let’s move on around. Mobility. Mobility with the advent of autonomous guided vehicles with the ability to move fixed cameras around within a location rapidly. Today I need to be able to see this. Tomorrow I need to be able to see with the same camera, a different angle. I need to be able to move this around. I need to be able to utilize a drone with a camera or a LIDAR to get to places where it’s unsafe for my workers to get, or impossible to get. I need to be able to use a robotic dog to get into a place in the environment that I can’t get to. I need to crawl into a location that’s radioactive that I don’t need to send a human in. So the utilization of autonomous guided vehicles to be a data acquisition method requires mobility.
Tony Lesley:
Move to data volume. Data volume is the amount of data that can be sent in a specific area. In cellular technology, there’s been a limit to how much data and there still is a limit, but the limit’s gone dramatically higher on how much in a specific sale, how much data throughput can be generated or transmitted, can be delivered. So data throughput becomes really important as you start to add more and more use cases and applications on these networks. We believe that the 5G network and 5G infrastructure that we’re talking about is not a one horse show. You’re going to be able to use this network for many applications of which computer vision is a really leading use case or technology.
Tony Lesley:
Device density. How many devices can connect in a particular location simultaneously? When we start talking about putting sensors on everything, everyone’s heard the promise of IOT and the number of sensors that have been dramatically increasing. And this has been happening for several years and the ramp is continuing to increase. So you have to have the ability to have many devices simultaneously connected in a specific cell or specific area. 5G dramatically 10X improvements on device density.
Tony Lesley:
And then the last, energy efficiency. This is done by several things, several technologies within 5G. There’s beam forming, the network knows where a device is. It can follow the device across a factory floor so that the device is not having to struggle to stay connected, lowering the requirement for energy consumption. Now, all of a sudden, you can put a sensor, a deploy sensor that might have a 3, 5, 7, or even 10 year lifespan in the field deployed. Who wants to deploy a battery sensor that you have to replace the battery once a year? That won’t be cost effective. So energy efficiency, another one that people kind of overlook is an incredibly important enablement or attribute for 5G especially in the factory floor.
Tony Lesley:
So the next question I always get is cool. Yeah. Great. Okay. 5G, that sounds wonderful but 5G doesn’t penetrate into my building. We’re going to go to the next slide here. And we’re going to talk about a new offering that Verizon has, where we’re deploying onsite 5G or private cellular deployments. We’ll work with you as a manufacturer, as a utility, as a city, as a venue. And we’ll deploy a private 5G network that you own. You own the network, you utilize our spectrum resources. You utilize the vast ecosystem of devices that are developed for the carrier’s public cellular network, but it’s your network.
Tony Lesley:
Now, the example I used earlier of 400 cameras doing machine learning computer vision in the manufacturer I was talking with earlier this week. Now I don’t have to have 400 rate plans with tremendous amounts of data utilization caps on those rate plans. It’s my network. I use as much as I want. Verizon just makes sure that it stays in place. It’s part of our network as a service offering strategy. Verizon has a new deployment strategy for network, our network as a service mass strategy. And this is one of the elements of that, this onsite LTE or onsite 5G deployment.
Tony Lesley:
So now we’re going to go to the other enabler that I talked about, the other rocket fuel that’s going to help us accelerate computer vision. That’s our MEC, Mobile Edge Compute or Edge Computing. So here, think about looking at the slide here, starting on the left. Think about your general purpose cloud for Edge. Today, that’s what we’re using. So your data from your devices is going to that kind of far away potentially, cloud instance or in your region but it’s going to that general purpose cloud. To lower the latency, one of the things that I talked about earlier, we’re going to bring that cloud closer to you. And it’s more than just latency, it’s also has the ability to lower network congestion, lower your cost of delivery, improve your data sovereignty so that you can keep things closer to you and or on your facility. So as we go from general purpose, what’s happening now is the Metro Edge. This is where Verizon has partnered up with the largest hyperscalers and we’re deploying their virtualized stack within our telco Edge.
Tony Lesley:
So now there’s 19 cities and this number is growing every day. There are 19 cities where AWS has their cloud compute and storage infrastructure built into the Verizon Edge. So now you can start to see where your devices that are out in the field can get to that compute resource and have that outcome come back to that device in a much faster, quicker, less data all the way through the network kind of instance. Then you move further on to the right. You look at the far Edge, this is a future. This is where we will be deploying that Edge Compute and storage out at our macro cell site. Think about a large macro cell site with Edge Compute there. Then you come further on to onsite. This is where we see onsite Edge coupled with that private cellular solution that I talked about earlier in a manufacturing facility, really paying off dividends especially for computer vision.
Tony Lesley:
So think about this, think about you are acquiring data, video sensor data, lots and lots of data. Some of that data because of data sovereignty, you don’t want that out onto the public cloud, or maybe it costs too much to track that data across the network, or you would like to track that out to the general purpose network so you can have your data there and you can do your AI modeling and training. You can adjust for data drift, the various things in the life cycle of computer vision. But you want to do that out in the general purpose cloud, where it’s a little bit more economic. You have your economics in play there. But you want that inferencing model to be very close, maybe even on site, maybe attached to your private 5G network. So now we’re talking about 5G network with all the attributes that I’ve discussed connecting to your virtualized compute stack from AWS or Microsoft to Azure, we’ve partnered with both, doing the actual inferencing right there on site.
Tony Lesley:
So now, these two enablers, these two rocket fuels are what we feel are going to really accelerate the utilization of computer vision in venues, in manufacturing facilities, in heavy industrial plants. So hopefully this all resonates. Hopefully this makes sense for you. If there’s any kind of questions or anything, this infinite vision summit that we’re a part of is a wonderful place to come and learn a little bit more about computer vision and the promise of computer vision. So hopefully we’ll hear from you soon. Thank you very much. Have a great day.
