Episode 322: Inside Panasonic Tough Command: Connected Vehicles for Emergency Response with Kjeld Lindsted

Episode notes

Most first responder vehicles are full of technology that does not talk to itself. A dash cam here. A radio there. An MDT, a license plate reader, lights and sirens, all from different makers, all siloed. The data exists. It just never lands in one place where anyone can use it.

Hello Smart Firefighting Community!

Welcome to another episode of covering real world innovations via interviews with fire service and technology industry experts that empower YOU to develop your very own Smart Firefighting strategy.

Kjeld Lindsted is the Associate Director of Software Products at Panasonic . His team built Tough Command, a software platform that wraps the Toughbook already sitting in several hundred thousand police and fire vehicles.

One device. One cell plan that already exists. Every piece of vehicle data pulled into the cloud one time, then made useful. Where the vehicle is. What its status is. What it is doing right now. Three simple pieces of information that underpin almost everything in connected emergency response.

This conversation gets into what that makes possible. A fleet manager who asks a question in plain English and gets an answer. A chief who gets a text the moment a vehicle crosses 80 miles an hour. A green light that clears the intersection before the truck arrives. And the harder questions underneath it all: Who owns the data? How does it stay secure? Where is the V2X mandate really headed?

Topics covered:

From Toughbook hardware to the Tough Command software platform

The three data points behind every connected vehicle

Asking fleet data questions in plain English with built in AI

Why a logged SQL query matters for security and compliance

Fleet managers, command staff, and IT in one platform

Real time alerts, from an 80 mph text to a high speed chase

Traffic signal preemption and the connected intersection

Geofencing for public policy and community presence

Data ownership, CJIS, and a zero trust security model

The V2X mandate explained, and where it goes in five years

So take a listen to hear what he has to share.

Head to www.smartfirefighting.com to discover how SFF accelerates innovation for emergency responders, to find out when our next event is, or review our curated resources.

Facebook | Instagram | Twitter | LinkedIn

Chapters

1. 0:00 Intro
2. 2:17 What Tough Command is
3. 6:21 Turning device data into value
4. 9:30 How the platform works
5. 15:18 Who benefits: fleet managers, drivers, the public
6. 19:49 The command intelligence platform
7. 25:22 Data security
8. 30:30 Vehicle-to-everything (V2X)
9. 37:19 Autonomy and where society is headed
10. 41:14 Words to live by

Questions answered in this episode

What is Panasonic Tough Command and what problem does it solve?

Kjeld Lindsted says Tough Command is a software platform that wraps around Toughbooks and other devices already in police and fire vehicles. It addresses the challenge that first responders face where a lot of their equipment is disconnected and from different manufacturers, bringing it together into one system.

Hear it at 2:03

How does Tough Command use AI in its reporting?

The platform embeds AI that lets users have a natural language conversation with their data instead of writing complex queries. Kjeld Lindsted demonstrates asking questions like show me a list of vehicles by mileage or add engine temperature, and the system interprets the request, writes a SQL query behind the scenes, and produces reports without needing special commands or technical knowledge.

Hear it at 9:30

What types of data does Tough Command collect from vehicles and devices?

The system pulls data from Toughbooks and other Windows devices, connects to vehicle systems through the OBD port to get engine temperature and odometer readings, and integrates with equipment like lights and siren systems, gun locks, plow blades, and utility vehicle arms. Kjeld Lindsted says it also connects to GPS, modems, and mobile printers.

Hear it at 6:51

Who owns the data that Tough Command collects?

Kjeld Lindsted states that all data collected belongs to the agency. Panasonic is a custodian hosting and making the data useful, but does not take ownership, sell it to third parties, or use it to train outside models. Agencies control which data elements they collect and how long they keep them.

Hear it at 25:51

How does Tough Command use geofencing for first responders?

Departments can draw geofence boundaries on a map to measure how often vehicles enter an area, dwell time, and crossings. Kjeld Lindsted says this helps with community policing policy and also powers traffic signal preemption by mapping intersection lanes so the system can request a green light as an emergency vehicle approaches.

Hear it at 22:59

Transcript

Machine transcript of the published audio. Names may be misspelled and some sections are missing.

Ep 322 transcript

Inside Panasonic Tough Command: Connected Vehicles for Emergency Response with Kjeld Lindsted

Speakers: Kevin Sofen, Kjeld Lindsted

Source: Whisper medium.en machine transcription of the published Podbean audio (no original transcript found in Drive), lightly cleaned. Speakers were separated by voice clustering and then checked and assigned by hand at every speaker change. Gaps where the model returned no usable text are marked [inaudible m:ss].

[0:00] Kevin Sofen: Kevin Sofen here where we cover all things related to the future of smart technology for first responders around the world. [inaudible 0:13]

[2:03] Kjeld Lindsted: to kind of add value to that investment that our customers are already making in the hardware side of it. So we're bringing a software platform to market that kind of wraps the Toughbook. One of the core use cases that we. and putting together our Toughbooks that are actually in police and fire

[2:17] Kevin Sofen: So we're looking at maybe one of the challenges that first responders have faces that a lot of stuff is sort of disconnected and, different manufacturers and a [inaudible 2:54] to be. What does what does look like and mean and practice from a hardware and software

[3:26] Kjeld Lindsted: and overall tough command perspective. There's a you know, if you think about the about the private market of vehicles, you know, the vehicles that you and I use [inaudible 3:39] And it's not because those things aren't useful, , it is just, how do you integrate them into police and fire workflows? And so [inaudible 4:21]

[6:21] Kevin Sofen: that could be valuable data [inaudible 6:23]

[6:51] Kjeld Lindsted: What that really means is you've got devices out in the world that produce some kind of a information, whether it's a sensor or some sort. And you want to pull all that back from a collection of devices. You want to land that in the Cloud so that you can analyze across that collection of devices. That's IoT in a nutshell. That's really what we do. There's kind of a lot that's involved in how do you talk to mobile devices that sometimes aren't turned on and sometimes are moving around in the world and you can't do the kinds of traditional network security or network testing that you might do with a fixed device. If you have a fixed endpoint, right, that's always on it's always supposed to be at that IP address. You can ping it and you know, you're paying it every five minutes or once a second or whatever your use case is and it responds and it says I'm on or I'm off and you can move forward from . If you have a let's call it a laptop that's in a police car and it's turned off sometimes, you can't reliably use ping, for example, to determine whether it's active, it's working, it's connected or disconnected. So there's a whole bunch of stuff that is sort of different about things that are moving around in the world that have this kind of transient property to them. Getting information off of those in a reliable, scalable, secure way is what my team is really an expert at. So to actually answer your question, yeah, we're happy to take data from sort of a lot of different sources. Today, you know, we're pulling it off of Toughbooks, but we've started to extend into really any Windows environment. We have on our roadmap building iOS and Android applications that pull in mobile devices. And then we're also plugging into things that are in the vehicle. So we plug into lights and siren systems. We can connect to a gun lock. We've done plow blades on snow plows, a salt spreader, the bucket arm on a utility vehicle. So at the edge, there's all kinds of information that can come in, but at the cloud level, we are also happy to pull in data. So we have two-way APIs where we work with partners like GIS platforms for situational awareness and other things like that, where information is necessarily going in two directions, depending on the use case.

[8:47] Kevin Sofen: That's awesome. And I continue to want to put myself in the shoes of a first responder now, thinking I'm, again, driving an SUV, and maybe it was upfitted four years ago, and you're continuing to maybe upgrade a few of these different, whether it's MDT or the radio, And again, this evolution of tough command is happening. Now, if you could maybe even visualize it or even screen any capacity of sort of like, what does this look like now in practice so again, I wake up and I go to the station tomorrow and how does my workflow, how is it different now and sort of from beginning to

[9:30] Kjeld Lindsted: You know, I love to talk about this stuff. I get excited. So I'll try to keep it really simple. So we do a couple of put in while vehicles are being out fit, but we can also install after, and then we equipment, is of interest to the customer. So, sometimes it's very often it's license siren systems, but again, you know, from that system USB into the back of cloud. Once the data lands in the cloud, we've got this great platform and this is where I'll share screen where we do analytics against that and there we go. So where we do analytics against all of that data. So keep in mind, you know, we've got this database that lives in the cloud and you've got, you know, typically roughly one second updates coming from a fleet of connected vehicles. So as you can imagine, you know, when we did some of this work in the past under other earlier iterations of the platform. We were working on D. O. T. P. A few hundred vehicles, and we would get billions of data points. So it scales very well, but it does scale quickly. And so we have a lot of information in the background. And what we're doing today is we're actually leveraging the power of A. I. built into the platform in a secure way that lets us actually mind that and kind of have a conversation with the data, which I think is really cutting edge compared to, you know, reporting tools that have been around in the past. Typically, date range, you'd maybe write some if then statements or something like that. You pick data elements and you get a report and then you sit there and stare at the graph and think, I don't know if this really answers the question I'm trying to ask or not. And maybe I'm not sure what I want to ask, right? So what we've done, we got all sort of the basic building blocks the table stakes things of a platform like this, right? User management. We have that I kind of want to show off with you today is this reporting tool. And so I'm just going to, we'll kind of a conversation with it. We're just going to play around. So I'm going to ask it, let's say I want to see, show me a list of vehicles by mileage. And there's no like commands or wake words or anything. This is kind of a real AI platform. We're built on Panasonic has a global AI team that puts some guardrails around the big commercial models makes them available to teams like mine to do building against. And so It's embedded right into the product, but it works the same way as you might use, you know, Anthropic or OpenAI products where you just have a conversation in natural language. So I have to, for a list of vehicles by mileage, you know, it's coming back with, okay, I hear you, here's a couple options of what I hear you, here's a couple options of what I think you might mean given the data that's available. So I'm gonna tell it, I want number one, that's the latest ometer reading. Now it's going to interpret that and what it's going to do behind the scenes is actually take that natural language input and it's going to create a SQL query, just like if we had, you know, a data engineer working on the team and we can actually pop this up and you can see there's a whole query that gets written. That's important because from a security and validation perspective, that actual SQL query gets logged. And so we have a full record of what the user asked for and what the system produced. So for those of you thinking about security and compliance, we have that record. It's an auditable trail. We can go back and see, what did it tell us? Was it accurate? Do we need to work on it? All those kinds of things. right, it produced this great table of data for right now. This is a sample vehicle, right, so this all this sort of internal testing data, but you can imagine this across a fleet of 500 vehicles or something like that in the department. So it's given me this, it's given me the answer, right? But now let's say I'm not done,. I want to play around with this a little bit more. So I'm going to say, you know, maybe the VIN is not something I'm familiar with. We use like a badge number, let's say for these vehicles I don't know which with which then it is so I'm just gonna tell remove that right so tell us remove the then what that thing for a second and see so it updated the table right I'm kind of in and then let's say okay I'm worried about these vehicles are making this up maybe they're there they've been overhe overheating right so we want to see at last none let's say engine temps to be specific you actually don't have to spell right and I mean would you be

[14:01] Kevin Sofen: able to ask you what are you could probably even ask it what are the what are all the data points that you have access to and then maybe even you're

[14:08] Kjeld Lindsted: you're reading my mind Kevin you're reading my mind yep so you can see here it added engine temp but exactly what I say so let's say okay I'm not sure what else I want to ask it so yes what other data and you show me and yes you can have kind of just a conversation with it and it's not always just for reporting right you can use it like people use traditional So here's a list of built this because early on we got feedback from customers. There's one big DOT customer that told us, that you can't build a button for all the things that I might ask. And we took that really seriously and we thought, okay, but AI can do that. Right. And so, you know, we can have this kind of conversation and you don't have to know what you come in looking for it, but you'd have to know, you have some problem, right? You're something you're trying to diagnose. I'm showing vehicle telemetry, but we also have information on interactions with geo fences. You know, is it in an area out of an area?

[15:18] Kevin Sofen: dwell time. What I'm seeing here, this seems like it could be, it would be very useful to a fleet manager to someone who's maybe kind of looking at, I mean, how many miles, what was the last temperature? Hey, we're seeing a consistent reporting compliance diagnostics. Yep. And then you were kind of just going into geo-to geofencing. That might be relevant to maybe less of a fleet manager of course it could still be them but also maybe again the battalion chief you know someone in command that's sort of kind of trying to figure out who's been where and again we're not trying to necessarily always play big brother as watching but yeah talk me through maybe from less of a fleet manager perspective which is extremely valuable use case but kind of a chief

[16:05] Kjeld Lindsted: if you will. So one is definitely the fleet manager, as you've identified, that's absolutely important to us. Whoever's responsible for all these vehicles, right? They've got a lot of work to do and giving them some intelligence is important. But the other two are command staff who are looking at safety functionality and then IT personnel who are sometimes trying to figure out which laptop and ticket printer ended up in which bin of the vehicle after the last one crashed or something like that, right? Things moving around, what's working or not working. And I think we have capabilities and kind of all three of those directions. So it's a little bit hard to see, you know, some of the, you can sort of see how we get access to the data here, but one thing that's really important, so I'll sort of show this just briefly. So let's say this is the perf report, right? This is exactly what I wanted to know regardless of which persona I'm playing. I can save that report. And then there's another part of the platform where I can go in and I can create automations that say, okay, you know, every Friday at three p.m. list to the shop, it's a list of trouble codes that are active or sort of the other side of that for command-staff, we had a police chief in the St. Louis Metro who told us, if one of my vehicles goes over 80 miles an hour, I want a text message on my phone because that means something's going down. And so, there's completely sort of different side of the data, it's built collection and reporting capability, but that becomes kind of a real-time alert that gets pushed out to that individual to say, it's maybe the first notification they get a fire is something that's going on that they want to get involved in, we might actually be the first place that they find out about it before there's a dispatch call or even a 911 call or something like that. There's a lot of power in different directions. Then if you extrapolate one layer further on that, what we're doing is now working with partners to take some of those alerts and actually push that information out to the even to the driving public and say, hey, you know, let's pop up on your navigation app and let's say, hey, there's a fire truck approaching from behind. You know, you need to be eyes up and pull over that kind so that you have to think sort of in concentric circles moving out and how does the data apply to the driver of the vehicle, command staff, the fleet manager, you know, the public that surrounds it, you know, that there

[18:20] Kevin Sofen: actual the firefighter, the police officer, that's the driver, and then also the public from this platform, and how is this whole ecosystem with Tough Command helping and applying, and how is that firefighter experiencing it, and maybe the police officer experiencing it. That'd be the first question. The second will maybe touch on the general public and where this kind of platform ties in.

[18:45] Kjeld Lindsted: There's a couple of things. So if you're the driver of the vehicle, It depends on how your department is set up. So some of these are cool vehicles, right? Where the driver takes that vehicle home, they're responsible for the maintenance and we're working with some, as we're deploying projects, some of the people that we're actually interfacing with are actually the end driver of the vehicle. They come in off shift we deploy their vehicle and intersections because we've implemented traffic-signal preemption and they may not be under the hood looking at this data and doing this kind of analysis, but they're definitely sort of feeling the safety and efficiency result of it in that their drive time is shorter, their response is shorter, there's less chance of a collision along the way. So I think, you know, it really depends. And the data is super valuable, I think, across the department. How you interact with it depends a little bit on your role and the kinds of things that you care about. But I think there's a huge case for just about everybody that we've run into in these departments so far.

[19:49] Kevin Sofen: I'd love to kind of continue to dive into this, the command intelligence platform we're looking at.

[19:54] Kjeld Lindsted: Yeah. So there's some cool things. So, you know, we've looked so far kind of at the vehicle side, you know, let's day. I'm going to go a different direction, right? Show me connected laptops. Now let's see what it does here. So I'm ad-libbing a little bit. There's not a ton of data in this particular demo platform. So let's see what it can give me, right? Just going to think about that for a second. There we go, right? So here's a couple of tough books that are in with a couple of pieces of information about connectivity, right. So you can imagine this is kind of the, let's flip from the vehicle side and let's talk a little bit about what happens in the IT world, right? I'm trying to diagnose, I don't know, I've got cell signals that are dropping in a particular part of my city. I've got devices that have failed and I'm not sure why, and I want to pull device temperature, I want to look at CPU utilization, trying to diagnose some app that's misbehaving in the background or something like that. So all that stuff that we were just talking through for vehicles, very much of that applies just as well to the IT world where I've got mobile printers that are connected, I've got GPS coming in from a cradle point or a Sierra modem that's attached. And I'm trying to understand kind of, you know, where are my devices and how are they performing in the real world? So same kinds of use cases, but now we're in the ITC instead of the fleet manager right? And a ton of power, all that data is coming into the platform as well. And you can see, you know, from the AI perspective, it was really easy for me to flip to asking questions about electronics, you know, where we were in a vehicle mode a minute ago. And I think that kind of shows that from my perspective, you know, we're building products that we want to answer questions and make data actionable for these first responders. And, you know, while I think in terms of buckets of different sort of user types, we really want the platform to be kind of self-service to give people whatever it is that they're in need of in a particular moment.

[21:43] Kevin Sofen: Again, that's one piece of the puzzle. There's what you talked about going to the OBD port, as well as these kind of connected laptops. But from now on, even kind of a little bit more of the thinking from that deck you sent me about all these other sort of high level nuances from the emergency vehicle preemption to emergency vehicle alerts and how this kind of ties into the macro smart cities conversation. just trying to make sense of it all. Again, we see these infographics, and obviously I think , everyone trusts your company, but it's still kind of like, how does that tie into city infrastructure? What actual hardware is on those poles talking to certain vehicles? And then I know we're seeing a lot from some other companies, kind of the vehicle, the vehicle communication and letting first pedestrians know when a first responder vehicles approaching and kind of giving them a soft notification and sort of like that is the experience of just knowing that, hey, be alert. There is a fire truck coming your way. Break that down a little bit about how this ties into the overall smart cities IOT experience to again, just help first versus community be safer. Yeah.

[22:59] Kjeld Lindsted: So there are a ton of use cases where those kind of crossovers happen. So it's kind of all, so we pull the same information up from the edge, right? And so, and then we have to think about how are we interfacing between the vehicle and some of its operating environment and one of the places that we do that is actually through what I'm showing here is the ability to create geofences. And so, you know, you can, you know, I could, you know, I could, you know, I could, you've all seen tools like this. But the point of this be able to use data to defend that. So this is an intersection between public policy, between politics, the actions of the department and then kind of what happens in the real world and it all intersects. And so one of the ways that we do that is through mapping, through the power of these kind of GIS platforms, we can actually let the department work with the department to draw in that area. So I'm going to bound in that disadvantaged community that I'm concerned about. And then I'm going to start measuring how many vehicles that enter that area in a month. How often are they there? What's the dwell time that they're in that community in the last two days? How many times do they cross in and out of that boundary zone? There's all sorts of analytics that you can do against that gives you real power to decide, hey, you know what, maybe we should have a better presence in that community or actually we really do a good job already and we want to be able to represent that well to counsel at public meetings to community members. So that's just one example of taking policy and the operation side and tying it into the real world. But there's another place. So you mentioned traffic signal preemption. So we actually use that same capability, and we can map in traffic intersections, right? And we map the input and output lanes of those intersections. And we've now created another interface point where we're not talking about policy anymore, we're talking about traffic signal preemption. And we can, using that same geofencing capability and some other mapping stuff that we see behind the scenes, we're able to actually say, okay, this is the ingress lanes of a particular traffic signal. As an emergency vehicle comes into this zone, we want to trigger a green light for them, right? And we work with partners on the DOT side to actually make that request to the traffic signal, but it's all premised on the same set of data, right? And so just a couple of kind of kind of completely different directionally, different examples of. existing infrastructure, with existing policy, and try to make the data work for the departments as we're pulling it in.

[25:22] Kevin Sofen: And of course, the ongoing question that is always talked about in the realm of data and AI is this security of the data? Obviously, your company's been around for quite some time and there's a level of trust with the brand but can you, from of a high level perspective, speak to how data is created, how data is transferred, and especially in the realm of AI and knowing that this isn't going out to train someone else's model with all the data that's being created. Speak to that a little bit if you can.

[25:51] Kjeld Lindsted: Yeah, I think that's an excellent question. And I love that people bring it up because I think people should be thinking about this kind of thing, right? We want to be building security first. So there's several directions I can answer that from. One that I think is really important for me to state is all of the data that we collect, it belongs to the agency. Panasonic is not taking ownership of that data. We're a custodian. We're hosting it and trying to make it useful for agencies. We're not taking it away. We're not selling it to third parties, that kind of thing. It's your data, you own it, you control which data elements you collect, which ones you don't want to collect and how long you want to keep them. That's all entirely up to the agency. And we have different agencies that have different policy reasons for, we have some that tell us, I don't ever want to collect speed because that creates a potential liability. And we have other ones that say, I absolutely want speed because it defends against the potential liability, right? And so it's not up to us to make those decisions. That's agency based and we respect that. We're here to facilitate. So, that's one thing. On the security side, we have a really robust security architecture. So we're working through certifications like SOC 2 and C just certification as we go. Now, we're building in a way that's as robust and sort of follows the best practices of any of the big kind of commercial grade deployments for security. So, you know, everything's encrypted. We have this kind of environment where there's not assumptions that once you're inside the platform, you can kind of do whatever you want. And we have checks and balances in there. And then we have thought through a number of different vectors of attack and how to defend against those things. And it really comes back to kind of a layered approach. So there's checks and balances in the system, there's encryption in the system, things that are sort of permissions based that like, you know, this particular entity is never allowed to do this thing because we know that's a particular attack vector, right? And so it's, it's a layered approach that includes things like monitoring of the platform. Do we see a spike somewhere or a drop off or something that indicates that you might be going wrong, policies around which users get added and which employees at Panasonic have access to these things and so forth, kind of on a need to know basis. And so we really take that seriously. and I'm happy to share with anyone we get into a project with, we can share kind of a, we have a document that we've put together that walks through various potential attacks scenarios, how we defend against them and how the platforms are to protect their data and keep them safe.

[28:01] Kevin Sofen: I mean, it's the old catch me if you can ongoing battle, it's cat and mouse,

[28:06] Kjeld Lindsted: so it always will be.

[28:07] Kevin Sofen: Yeah, no matter how great your continued cybersecurity and security hardening will be, there will always be the cat chasing you

[28:17] Kjeld Lindsted: So we certainly try. I mean, another good example I should have mentioned in that kind of stack is we actually use the trusted compute module that's built into the laptops that we touch to validate that endpoint. So it's impossible to kind of insert of insert a kind of a man-in-the-middle attack. So there's a lot of different things from the hardware level all the way up to, you know, encryption at the data storage level in the cloud and everything in between. Because like you said, people are always poking around. we know that we're realistic about how the world works and we're always on top of it.

[28:45] Kevin Sofen: I mean, I've kind of been dabbling into the hacking world and more so just playing around with vibe coding and AI agent development. And then it is, and also you're hearing about the new anthropic mythos drop, you know, having a model that is finding gaps that have been hardened for 30 plus years in the internet with thousands of people who have worked on it, kind of exposed overnight. you're continuing, do the best you can. So I appreciate that context.

[29:13] Kjeld Lindsted: That's actually a really interesting example. I'm going to give a shout out to my engineering team. The way that they work is actually pretty cutting edge too. So they're actually using some of these latest tools to do things like code review. And I think that's really important because as you said that some of these AI tools are now poking holes in sort of standard practice for software development in some cases. We're trying to stay ahead of that so not only making sure we follow industry best practices for how the platform is architected but making sure we use the latest tools as we're building to check ourselves to [inaudible 29:43]

[30:00] Kevin Sofen: if you're sort of just looking at these tools and saying, oh, we'll never use that's not real software coding. I think that's a recipe for failure. But if you are a software team that knows software and is leveraging these new tools to build more software and use it for code reviews, I think you kind of got an unstoppable combo there.

[30:16] Kjeld Lindsted: The example that you made of, you know, finding holes in stuff. So we do things like that, right? We run our own internal audits. I mean, I think that's important. There are some really good uses for some of these tools that kind of help keep us safe in the process of developing through it.

[30:30] Kevin Sofen: Yeah, and one thing that, frankly, truly mentioned in the intro, haven't touched on yet, but the whole USDOT V2X mandate, for those that aren't familiar with it, obviously that's kind of a big deal with the whole vehicle to everything concept and what that means. Like, from a high level, what is the V2X mandate, and how does this kind of all fit into the mix of that whole new mandate coming out from DOT?

[30:55] Kjeld Lindsted: Yeah, that's a great question. So there's been several iterations of sort of this V to X, which just means vehicle to everything, right? It's this concept of, how do vehicles broadcast safety specific messages in a context where they can be received either by the infrastructure, so roadside operations or by other vehicles and use cases for that are things like if I'm behind a semi-truck and the car in front of them slams on the brakes, I might not be able to see that, but I could get an electronic message automatically from the computer of that front car and the semi-truck, both broadcasting back saying, hey, we're hard braking, here's our geometry and locations. My car's computer could, in theory, ingest that information, actually trigger the braking before I, as a human, have a chance to respond to that incident. I may not have even seen it happening yet. So that's kind of the underlying premise of VDEX. Now VDEX as a technology has been around, the early iterations of it were kind of in the mid to late 90s. And one of the challenges with VDEX has always been, and my team sort of started working on this inside Panasonic for several years, But it was always that it required upfitting vehicles with pretty significant, expensive new hardware. And so what my team has done over the last few years is those are very valid use cases. They haven't scaled the way that we would have liked in the industry for a number of reasons, but cost being kind of a really big one there. And my team spent a lot of years working on some really big DOT projects. We're really proud of the progress there. But what we've done recently is take the core principles of that kind of safety messaging between vehicles and transport that into a place where we've got an existing hardware footprint in the Toughbook where we can leverage, we can't do everything that original VDEX world imagined, but we can do a lot of it and we can do it without having to add new hardware to the vehicles. For us, that was a, let's not die on the hill of, we have to use this particular hardware. Let's see how we can get closer to the safety applications using stuff that's already out there. That's where we are today. Of course, I think there's a lot of safety benefits to the VDEX paradigm, if you will, and I'd love to see it moving forward. And we're trying to do what we can with the footprint that we have available.

[32:56] Kevin Sofen: And I mean, I just was doing a little bit of research earlier and just this, I mean, the ultimate goal is how to reduce death and serious injuries on America's roadways. And I mean, obviously that's sort of like this grandiose goal and you forget, was it project? Maybe it was project 2020, it was going to have to have like zero pedestrian to pedestrian deaths. And I think objectively it's all, a lot of these sort of government, federal mandates and whatnot, they sound good, but then obviously kind of the practicality, the implementation of industry is just a bit harder than when really trying to make it work from the equipment within the vehicle, the vehicles themselves, and actually kind of the training and onboarding. So I'm just kind of curious in where it's at today and maybe your Kel division in five years, and of course kind of the role do you see? I hope to see AI is an enabler for this. But yeah, where do you see this whole V2X mandate in five years and where Panasonic plays in that? Yeah, it's an excellent question. I definitely have some sort of personal opinions

[34:15] Kjeld Lindsted: working on traffic signal preemption and some other things and some of the older engineers in the department kind of jokingly called me the flying car guy because I was talking about, you know, self-driving vehicles and autonomy and AI and all this stuff. And this was, you know, 2012, 2015 kind of early stage for some of this stuff. So I've been poking around these things for a long time now and kind of, you know, I'm a sci-fi nerd and I love kind of, you know, the future might be possible has always intrigued me and I think we're in this really interesting space so that the VDAC technology place started in the late 90s when you didn't have 5G. I don't know what they had 2G back then like, you know cell phones were themselves still early. There was no iPhone, there was no compute in these vehicles, there was very little connectivity and it had limited bandwidth. So it makes sense in that world that you've got to put a dedicated radio to process this kind of safety communication. happened is, and you're exactly right. So there was an early an early mandate. But there are two things in the industry that have really kind of challenged the validity of the mandate, I think from a development perspective, the first one is communications just gotten so much faster and so much more ubiquitous. I mean, all of us today are carrying around a , pretty powerful computer with very high bandwidth, very fast, you know, very low latency, fast communications. And so, the need for a dedicated communication is just, I think it doesn't exist the [inaudible 35:44] really high quality I think we will see a significant drop in roadway deaths, but I couldn't say for sure if it's because of autonomy, or communication, or VDX, or

[37:19] Kevin Sofen: or VDEX. and he pressed a button to go there and the car took us there. So I was blown away. And he takes it like an hour long commute to work And even that, he was saying he kind of just sits back and relaxes. And he told me the one restriction is you can't put anything in your steering wheel and letting the AI and the software running it for you. And then of course, once that first accident happens, yeah, then it's like, oh my gosh. And then you look at the amount of human accidents. So it's, again, there's a cost-benefit risk of all these technologies and humans involved in the mix.

[38:02] Kjeld Lindsted: And we're at this weird intersection, right, where Tesla's got some stats and I'm not here to pitch Tesla, I don't work for them, but they've got some stats about the difference in severe accidents between the self-driving system and sort of average human and they're pretty compelling stats from what I've seen. And I know there's some arguments to sort of how they did the measurements and things like that. There's always gonna do to be a debate. I think that's healthy, but I do use it. I pay attention while it's running, but the number of interventions that make sort of drops off on a pretty consistent pattern over the year or so that I've had it. And again, it's not, it's not to the point where I think, you know, I would go to sleep in the car, but you can kind of see, again, I'm a sci-fi nerd. So I think here's where we are, here's sort of where, you know, Star Trek or Star Wars or whatever might be with these kinds of technologies. So we kind of know what theoretically might be possible. And I like to look at how are the technologies we have today kind of stepping us towards that sort of theoretically possible future. I think you can sort of see it coming in some of the autonomy that's happening now. And that's pretty exciting to me.

[39:02] Kevin Sofen: I'm gonna watch some Black Mirror episodes tonight in honor of this conversation.

[39:06] Kjeld Lindsted: There's definitely a big potential downside I think as a society, we have to be really careful about that, but it doesn't mean the technologies aren valid. It just means we have to sort of make sure we do them the right way.

[39:15] Kevin Sofen: Yeah. Do you ever watch Black Mirror?

[39:17] Kjeld Lindsted: I haven't, no. So, and not for any good reason, I've heard it's good. I like to sort of think about the sort of positive side of technology. And I don't think I'm sort of head in the sand, but I'm less excited about sort of how things go wrong and more excited about where are we going to get it right.

[39:31] Kevin Sofen: Yeah, no, I think it's important to watch Black Mirror to kind of see sometimes to see the pessimistic angle but also continue to think about the opportunities and the good but there and some of the aspects of how the idea of skill. and just kind of like the idea in the matrix when Neo learns how to do Kung Fu over, you know, kind of use programs and all of a sudden he's the best Kung Fu fighter. So, excuse me, Morpheus. And it's just kind of crazy to see how the software is continuing to advance. But like you said, I want to continue to see it as a force for good. Of course, there's a lot of like AI doomsday mindset out there, but how can we continue to leverage this as a force for good for first responders, for society, for humanity, and you and your team are very much on the cutting edge of it. So thank you for the. for the good work you're doing.

[40:36] Kjeld Lindsted: We try to be. Yeah, and so real quick on that point, I think there's a really interesting quote. There's a financial podcast that I listened to. So not related to technology, but they were saying recently, there's some economic turmoil that there always is, right? And one of the things they were saying is a lot of the reasons that these things tend to work out is there is this vast conspiracy of some 8 billion humans who wake up every day and say, how can I make this. How can I make the world a better place? How can I make this work? And there are obviously bad actors out there, but most of us And we try to make things better. And I think that's generally why sort of the human species, if you will, continues to survive and thrive. And I like to think that my team plays a small part in that with the work that we

[41:14] Kevin Sofen: do. Heck yeah. I was actually going to ask you if you have any kind of quotes or words of wisdom you live by. You kind of just dropped it there. But maybe even a final prompt.

[41:22] Kjeld Lindsted: Are there any other- Vast conspiracy of success.

[41:24] Kevin Sofen: I love it. Vast conspiracy of success Vast conspiracy of success. I'm going to paint that on the wall here.

[41:28] Kjeld Lindsted: I thought it was great when I heard it too. I think that kind of encapsulates my personal philosophy There's all sorts of clouds on the horizon. But you know, I've survived for all these years so far and things keep getting better in a lot of ways and I like to look forward to that.

[41:41] Kevin Sofen: Right on man. Well, excited to see some of your team at FDIC. We're recording this today. It's April 14th, this will come out in a little bit. But thank you for the good work you and your team continue to do. Excited to continue collaboration and until

[41:54] Kjeld Lindsted: next time. Yeah, fantastic talking to you, Kevin. Really appreciate you bringing me on. It's been a pleasure. Right on.

[41:58] Kevin Sofen: Take care, man. Thank you so much for listening to the Smart Firefighting podcast today. If you enjoyed what you heard and got any value, please drop us a rating, leave us a comment or reach out to us on social media. Have a great day and together we can advance the future of Smart Firefighting.