Innovate and modernize connected vehicle platforms with AWS IoT

Session IoT 204, Innovating and Building and Modernizing Connected Vehicle Platforms with AWS IoT

Now, for those of you in the audience that are building or enhancing or designing a next generation connected vehicle platform for your company - got great news for you. You are absolutely in the right place. This is a session that's focused on the ways that we help you innovate around your connected vehicle platforms and we have a lot of ground to cover.

We have reference architecture, stories, great content, a lot of ground. So we're gonna jump right into it.

So really fast for speaker intros, my name is Larry Snow. I am the Principal Worldwide Go-To-Market Lead for Automotive and Connected Vehicle for our IoT Services. Joining me today will be Andrea Ketzer. She's the Director of Technology Strategy for Connected Mobility for the Automotive Business Unit. And then last but not least we have a very special guest. Yozo Takehara will be joining us from Honda Motor Company for American Honda Motor Company, the Unit Lead for the Digital Services Development Division.

And with that real fast overview of the agenda:

• I will be covering Connected Vehicle Platforms with AWS and the three areas of innovation we can help you with related to our IoT Services.
• Next, Andre will cover Modernizing the Connected or Modernizing the Vehicle Data Platform.
• And then Yozo will be talking about his firsthand experience helping Honda migrate their platform to AWS IoT.
• And last, there will be some Q&A afterwards, but we'll shut down the session and then we'll, we'll wait over here. So if you have any questions, feel free to approach us and, and let us know.

So every good session starts with a story. This session goes, or this story goes all the way back to 1885 and the Benz Patent Motor Wagon. This is widely regarded as the, the first popular modern automobile. There's a rich story about how this came about, but the part of the story I like best is the inventor's wife Bertha Benz. She's famous for two things:

First, she took the world's first road trip in an automobile traveling 60 miles to visit her mother, which says a lot about her courageous spirit and personality.

And second, while she was on this trip, she discovered that there was a problem with the brakes. So she stopped in and worked with a local cobbler and invented the world's first replacement leather brake pads, an iteration of which we are still using today.

So why do I tell the story?

First to highlight an amazing woman and pioneer in the automotive industry. And that this is the beginning of a lineage of individuals and women like Andrea Ketzer that are still doing awesome things in the automotive industry.

And second, because adding innovations to vehicles is nothing new. This has been happening since 1885 and it continues today and it is the cloud that enables all kinds of interesting new use cases that we'll be going over today.

So what is it exactly that the cloud enables? At AWS, we're helping OEMs around the world build a spectrum of new features and capabilities that leverage cloud hyperscale technology and infrastructure, things like ADAS and autonomous vehicles, EV optimizations, battery health monitoring, map and trip services. You have connected grid and infrastructure over the air updates, subscription management. ML integrations - all of these things and many more are getting built and delivered better, faster, more securely and easier leveraging the AWS cloud.

And why is it that companies are choosing AWS to build their connected vehicle platforms? There are a number of reasons but these are the top reasons that OEMs cite:

1. Scalability - the ability to deploy 1 million or 25 million vehicles on the infrastructure that is literally ready to go today. This is one of the top reasons that we see companies migrate from third party solutions over to IoT because of issues they're having around scalability and resiliency.

2. Multi-region - the ability to deploy those 25 million vehicles in any region around the world that's important to you, including China.

And last but not least, cost savings - we've worked with customers that have migrated their message broker to IoT Core for MQTT that are saving more than a million dollars annually.

So for all of those reasons, many more, we now have more than 270 million devices that are connecting into the cloud with IoT Core. And many of those are vehicles from OEMs, tier ones, telematic solution providers and others in the industry that rely on the IoT Services for their connected vehicle platforms.

Now as context for the next set of slides, I want to go over how AWS views a connected vehicle architecture. So at the bottom, we have a devices and hardware layer - ECUs, TCUs, gateways and vehicle networks that need to communicate with the cloud back end. They would do this through the communication layer, which is predominantly MQTT. On top of that, you have a vehicle data platform, you have your, your, your software at the edge and then you have an operations layer that allows you to manage that fleet of vehicles. And most importantly, on top of all of that, you have the applications and use cases that ultimately they are the reason that you build and deploy a connected vehicle platform.

So to deliver the architecture that I just highlighted in the previous slide you basically have two options:

Option number one, you can build your own platform or option two, you can buy something or integrate a module that's that's complete.

Now for the builders out there, as you know at AWS, we love working with builders. We provide a rich set of tools to help you from infrastructure IoT services to provide those foundational building blocks to connect the vehicle reference architectures and blueprints to a vibrant partner community with, with partners like Bosch and SiriusXM and Connected Mobility Solutions that are available in our Solutions Library.

Now, for the other option, if you're in the buy category, you can work with great partners and solution providers that we work with, that have built and delivered their their solutions with the AWS IoT services underneath - that's partners like WirelessCar for OEMs, Motive for fleet operators, Vistion for OTA and Continental for EVaaS solutions. Whichever path you choose, we definitely still love you and we are happy to help you integrate your connected vehicle platform with your AWS environment.

Now, for those builders out there, this is definitely the focus of the session, the majority of the majority of OEMs that we work with already have a connected vehicle platform in place. They're already managing millions of vehicles. And so they come to us with a very important question. They say AWS, what can you do to help me enhance or modernize or add innovations to my existing vehicle platform? And we actually have a very simple answer.

We have three services that are available today to help you modernize and do awesome things with your connected vehicle platform. Those are:

1. IoT Core for managed MQTT
2. FleetWise for vehicle data collection and management
3. Kinesis Video Streams for vehicle camera use cases

And that's what I'm gonna focus on today. So think back to the architecture slide that I shared previously, this is what it would look like to integrate the AWS IoT services into your vehicle architecture:

You have IoT Core here that would help with vehicle communication, FleetWise would improve your vehicle data platform and Kinesis would provide vehicle live streaming capabilities. And with these infrastructure enhancements in place, you can easily build and deploy a multitude of use cases like remote vehicle commands or silent test mode for ADAS or vehicle 3D view and many others that are, that are available and we'll go over some of these.

Now, first, we're going to start with the first infrastructure enhancement - IoT Core for managed MQTT. So virtually every OEM and others we work with today have adopted MQTT or migrated from MQTT or building with MQTT or doing something with MQTT in some fashion. And that's because MQTT is the de facto standard for vehicle to cloud communication. The reason for this, it's low bandwidth low latency and has a published subscribed model.

So by migrating the message broker inside your platform to IoT Core for managing MQTT not only do you get the most reliable, secure cost effective scalable service, but you also make it easier to deliver new experiences to your end customers.

So this is a great example, a use case of one of those design patterns that we see many OEMs and others build using IoT Core. It is relatively simple, straightforward to enable remote vehicle commands. And this is used to enable start stop, vehicle start, stop or door lock, unlock, cabin preconditioning for heat and AC and many other things.

Now to implement this design feature, you have a companion application that runs on a smartphone or, or a mobile application that sends a command request into a Kafka queue via API Gateway. From there, you have a Lambda or an AppSync Lambda that reads from the queue and then updates the commands database or the commands table in DynamoDB. You then, if the vehicle status shows connected, which is also tracked in, in DynamoDB, a command is sent via IoT Core and a request response topic to the vehicle. The vehicle then implements the command, sends a response back via MQTT and that finalizes the command.

Now, I know that was a very quick overview of, of, of this design pattern. Just know that later in the session, Yozo will describe exactly how Honda migrated to IoT Core and implemented this very functionality in their Connect platform.

So next infrastructure enhancement number two, following the migration to IoT Core for managed MQTT. The next logical step is for you to examine ways that you can improve your vehicle data platform. That vehicle data platform is composed of two parts - you have the vehicle data collection that's happening in the vehicle and then you have a cloud back end for data routing and data management. This is precisely why we engineered and launched AWS IoT FleetWise, which is a fully managed service that we developed specifically for the automotive industry. Unlike anything out in the field today that makes it easier for OEMs and others to model, collect and, and manage to model, collect and manage their vehicle data all from within your cloud environment.

So we don't have a lot of time to go over all of the awesome capabilities of FleetWise in the session today, but know that we continue to innovate and add new capabilities based on customer feedback and we launched some awesome new FleetWise capabilities during re:Invent this week. And Andre will talk about these features in the next section.

So let's look at another design pattern that leverages FleetWise to build a vehicle health monitoring application. So in this scenario, the OEM has integrated FleetWise into their vehicle stack which you see on the left hand side, that would be based on the open source reference implementation of the FleetWise edge agent using the data campaign construct. The FleetWise agent would ingest data from the CAN bus or vehicle networks or video systems and transfer that data into S3 or Timestream based on the events, rules and conditions that you define within your cloud environment.

For the external applications on the mobile client, a request is made from the mobile application to see the health of the vehicle, the status of the vehicle. That request goes to the cloud via API Gateway. The data is then captured from the appropriate data store and sent back to the the mobile application for visualization. For internal applications, most customers are using managed Grafana or QuickSight to build dashboards and for analytics and insights and other applications for internal use.

And this is just one example of many and there are even many more advanced use cases that can be easily built and deployed by integrating FleetWise into your connected architecture.

Now, the third way to improve your connected platform is to integrate Kinesis Video into your platform for vehicle camera use cases.

So what is Kinesis Video Streams? Kinesis Video is a fully managed service that provides real-time streaming of video and audio into the cloud for storage, playback and analytics which can then be used to build out a range of new interesting innovative applications with that capability.

So with the increasing prevalence of vehicle cameras, both factory installed and aftermarket, it makes sense that OEMs and others are rushing to build new solutions leveraging these rich sensor devices - things like driver safety monitoring, remote teleoperations, 3D view and of course live streaming, which we will go over in the last design pattern that I will cover.

So with the integration of the Kinesis client into your vehicle architecture, you can easily build and deploy what we call vehicle security mode. It's a feature that would be available and this is how it works:

You have a car that's parked, that car is using sensors, accelerometers and other things to monitor activity around the vehicle. When an event occurs...

IoT Core sends an alert to a user's smartphone application. The vehicle owner sees an alert and opens up the vehicle, the the vehicle companion application and clicks to see a live vehicle camera stream that's in the vehicle. If the the vehicle owner sees something and wants to take a certain action, they can initiate a command from their smartphone. IoT Core would send that command back of the vehicle. It might be something simple like honk the horn or turn on the lights or do anything. Those are just simple examples, but that's how you, that's how the vehicle security mode feature would work.

Now for the architecture, you have IoT Core inside the vehicle that enables push notifications to the mobile application, mobile clients, you then have uh an IoT Core also sends the commands back from the mobile clients into the vehicle. Uh for mobile client communication, you have CloudFront, Cognito and API Gateway. And most importantly, you have Kinesis Video Streams with WebRTC for uh low latency live camera caption and and and peer to peer uh peer to peer communication.

So this is just one example of the many things that you can build with integration of Kinesis into your connected architecture. So for those of you that are interested in more of a zoomed out perspective, this is one example of many that we have for a complete connected vehicle reference architecture or blueprint. And uh we have, we have many of these that we share and work with customers is they're building out their connected architectures or modernizing their connected architectures.

Now, unfortunately, we don't have time to step through all of this today, but know that there are a number of sessions that dive into this later today and or later this week actually, and we will share those session IDs with you uh at the end of our, our presentation. And not only that if you would like to work with AWS on your vehicle architectures or as you design the next generation architecture, please work with your account teams to schedule meetings to include our connected, our connected vehicle, domain experts.

And finally, we have worked with many OEMs, tier ones, telematics providers and others that have done just this done just that worked with us to build out their connected architecture or, or to modernize their connected architectures. And these are just a few companies like Toyota, Hyundai, Continental, LG, Viston and notably Honda, who you'll be hearing about later today or you hear you'll be hearing from later today.

And with that, Andrea will talk about modernizing the vehicle data platform.

Thank you very much, Laurie.

Hi, everyone. I hope you can hear me. My microphone is ok. So let's continue. Let's get started with the puzzle we see here. Nothing more than hundreds of pieces floating around, unstructured and unintegrated. But if you look exactly of what that is and if these pieces are integrated in an, in a proper way, if they are assembled together in a proper way, the result is a wonderful car. But what we all cannot see on this picture is the most difficult piece of the car, the software. And the thing about the software is that it generates day by day, more and more data. And this is uh what I'm here to talk to you about today.

My name is Andrea. I'm in charge of the tech strategy for auto and manufacturing at Amazon Web Services. Before I joined AWS, I was in the auto industry in charge of uh product developments like surround view systems or top fuel v whichever term you are using, I was developing um central electronic control units for autonomous mobility for a a um complex radar and camera systems um at that time. Um but all these products, they are very different, of course, but they have something in common and this is complexity. Most of the time, most of my time in the auto industry. Sorry. Um we didn't have the cloud so we had all our data was on premise. We spend a lot of time to um to develop services to manage our infrastructure to free up memory just to have space for the new data. So I spent a lot of time with my team to find new ways, new tools and new methods to um to help them to focus on the product development itself and have less undifferentiated heavy lifting.

In the next 20 hours. I will be talking to you about how cloud can help develop um can help you accelerate uh development. Ok. Don't be worried. It's not 20 hours. Laurie gave me 20 minutes. So let's get started.

Data is as I said before, is one of the key elements of the car. If the data is delayed, the development is delayed and if the development is delayed, the next thing which happens is what none of us wants to hear the product launch is delayed, the start of the product, the start of production is delayed and that's something which we all want to avoid.

Today at AWS. Our goal is to help you accelerate the development over the last few years through a deep collaboration with industry leaders uh around the world. Uh we learned about the vision and the goals and the challenges the auto industry is facing um on their journey towards software defined uh vehicles.

We have heard from automotive customers time and time again that when it comes to SDV to software defined vehicles, the most important goals are:

Number one is to accelerate software and hardware development to accelerate innovation and time to market while continuously decreasing the costs.

Number two is the absolute desire um and expectation to increase the code quality, the software quality by uh increasing the test depth.

Number three is uh need to deliver more personalized user experiences.

And number four, last but not least is also the desire to continuously upgrade the vehicle uh through the lifetime of the vehicle.

One of the tools we have just released to help you accelerate your um development cycles is the Connected Mobility Solutions. The design of CMS or Connected Mobility Solution is our answer to the focus on developer experience at customers like Toyota Delante and Honda. CMS is transformational in the way how uh developers will innovate and deliver valuable experiences leveraging in-house development and third party offerings.

CMS includes dashboards uh which help you visualize incurred, expansive on services or on project or on both so that you have the possibility to continuously monitor, adjust and optimize your architecture to save cost by providing governance and control of deployment. Development teams can now innovate and test their new applications without impact to the consumer experience.

At AWS, we offer manage services and also solutions, AWS solutions and partner solutions for connected mobility that will help you reduce the costs of development and maintenance of your connected mobility platforms. They will also help you optimizing the data in just so that you reduce your your in just costs. And we help you develop the data strategy which uh helps you deliver more personalized um experiences to your end customers.

We also offer fully managed services that help accelerate the development. In 2022. If you have been here at re invent, you were maybe uh in the in the session where we launched a new fully managed service um AWS IoT FleetWise. So that was last year, one year ago. This service helps you automate company, helps you um help automotive companies collect vehicle data and transfer it to the cloud in a very efficient way. It allows you to model your vehicle data and standardize it so that it is easier for the next downstream uh analytics and applications.

Fleet device allows you to collect only the data which you really need for the use case, which you are working on. For example, if you are working on an on an emergency brake assist feature, for example, uh then uh FleetWise will automatically transfer uh data which is related to events relevant for that particular feature like in this case, break events. So the data is immediately available in the cloud. The developers can immediately access it. They don't need to wait until the test driver. The test engineer comes out from the car delivers the physical hard drive and uploads it and uploads the data. You save hundreds and hundreds of hours of waiting for your data. And of course also say you save time but but by looking at data which is not relevant for your use case.

In the world of data collection, we see increasing challenges um with a massive volume of data coming from cameras, data that we need to power autonomous mobility um and its machine learning, a modern vehicle. I'm sure you know that better has up to 10 cameras. It has four surround view cameras, it has a front looking camera, it has a rear view camera. Um it has a driver monitoring and many more. The number will just increase in the next year.

While automotive companies have been collecting vehicle data from vehicle sensors like speed or um fuel temperature or fuel pressure. Um it is becoming more and more important in the software defined era to collect also other data from other sensors in the car such as camera radar and leader.

Vision systems generate a massive amount of data in structures, structured and unstructured formats. The data often lacks context and it is difficult to synchronize making it challenges to assemble a full picture which you need to to develop your use case. For example, if you want to analyze um the accuracy of road conditions detected by the vehicle camera, a data scientist would want to analyze camera data um and some other vehicle data like for example, the speed and the brake pressure in that moment, keeping all of those data which are relevant for that particular use case synchronized um and associated with the same event. It's challenging and it's, it's a heavy lifting which is typically requires additional software to be developed and it requires a lot of compute uh power and time uh to keep the data synchronized.

And now I'm super excited. I have to have the pleasure to introduce you a brand new feature of the AWS IoT FleetWise service, which is our vision system data. So from now on uh you can uh collect easily, collect and organize data from vehicle vision systems uh that include cameras, radars and leaders. It keeps vision system data metadata and also other uh vehicle sensor data, telemetry data um synchronized in the cloud, making it easier for customers, for developers. data, scientists test engineers to assemble a full picture view of events and gain insights.

Customers can get started on AWS by defining and modeling a vehicle vision sensors alongside its attributes and standard sensors, the customers edge agents. So there is um the the service has a tiny agent which is running in the vehicle and it is trained to to collect only the data which you as a customer, as a developer you specify. And in the cloud, you will see the full picture view of all the data which you need. Your telemetry data, your vision data, the metadata um related to to this all in one place and it stays automatically synchronized, making it easier to use downstream for analytics and applications.

Continental was one of the first customers who raised our attention towards this challenge in the auto industry, silent testing or or shadow testing. I'm sure this is a term you are familiar with in in in the auto industry. So Conti came to us and asked for help for a co innovation uh to develop these services. So we want to say thank you for the excellent collaboration we had in the past couple of months and years and we wish you uh Continental and also many many other customers success with FleetWise. Um enjoy working with this service and contact us if you need more help. Uh we are happy to jump in and support.

And now my 20 hours are over. So I'm happy to hand over to Yozo to talk about the customer journey.

Thank you.

Hello, good afternoon. Uh I'm Yozo Take and the lead, the mobile app and the backend server for the digital services and development division on the American Honda Motor. So I love this picture. So the Honda Motor is pounded in 1959 and it's a fast overseas subsidiary of the Honda Motor Company.

"And we are responsible for the distribution marketing and the sales for the, the Honda branded automobile motorcycle power equipment and marine engine and jet. As you may also know, we participate in many motor sports including Formula One.

Additionally, our division, Digital Services and Development Division are responsible for developing operating the handling services for connected co-owner today.

Now, i wanna, i would like to talk about the actual state of the MT TT used in the market car. And first i will talk about the legacy architecture before we use the MQ TT and the few issues then. So i will introduce the actual usage of the Q and we will also talk about some obstacle we faced and how we dissolve the situation.

Finally, i want to show the actual services using the MQ TT in the market. This is a traditional connected to our ar architecture. So let me give you a brief introduction. So it's starting from the right edge mobile app. france's customer wanna do the remote operation, the mobile app, request the server to do the demote operation, demote command and the server side will uh process the command quickly and ask the network carrier which car connected to wake up the vehicle. At that time, the s ms like a text message we used to wake up the vehicle.

Then after wake of the vehicle, the vehicle will ask server to retrieve the actual remote command and execute. Finally, after execute the command, the result will send back to the server side and the server will notify to mobile the result. Then finally, customer will get the remote command result. But it's so long and complicated architecture. But the main point is so even we are in the internet wall. So we are still relying on the s ms technology that a legacy i wanna show the two major issue is this approach.

So first is the scalability. So we are increasing the car and also at the same time, mobile app also increasing and this backend server can scalable. However, the s ms is out of our scalability control. So even if the the our partner carrier uh including the service, but uh sort of by design. So during the spike, we will be a delay in the operation.

The second is a user experience. So it often happened, the device cannot get a remote control. And this is the major characteristic for the this kind of iod device and its move. So that's a server or mobile had no way to know in this situation before. So that's so the in the remote operation, mobile have no choice but just wait the response then. So finally time out, the customer will get the time out message after 30 to 60 seconds. So this is a bad user experience. we think.

Meanwhile, a announced the managed am qt as a aws iot in 2015. it's doing with my story. but uh i was developing a mobile in the honda motor in japan. i thought maybe this solution will work. so i came up with this idea and i created a quick p one is the app that simulate the car and another is the remote app. and i was able to quickly set up the server using the a iod console and it works fine at that time, it ended in the in-house demonstration. but after that, i was selected to develop the new and next generation car. so i was able to use this experience to introduce the aws iot to our car.

i hope that all of the you are interested in the new technology and they will take advantage of the opportunity like this with the big picture perspective. but so new technology often come with the risks. so we took approach to minimize the risks and maximize the use of the legacy assets.

so the actually the sns part is where we mq tt used actually used. so this approach is very simple change but works well. so we are able to get big benefit with simple change such as the so being able to wake up the car with low latency on the scale during the spike without dropping and the the ability of the whole man to connect it to mqdt server. and knowing that in other ones with the car is uh uh cannot accept the remote control.

additionally, aws has a wealth of logging and the monitoring to such as a cloud watch. so which was very helpful during the development.

now, i wanna show the one of the obstacle we faced and how we resolve the situation. it was occurred when the stability in the non production to the production environment. we initially tried to manage the non port and product in different aws account that this kind of independent environment like this would be easier to manage. we think.

however, later we found out that the since due to the aws iot restriction, so same certificate cannot be used in the two account. so that means so it will be necessary to replace the hardware, depend on the server environment. so we thought this is major block for the development speed.

therefore, the solution we took was to manage a iod non proud and proud in the same account but separate the the environment by topic, this allowed to uh switch environment without the hardware replacement. so it works fine.

lastly. so i wanna uh introduce the honda car that utilize the aws iot. this is a new accord. compatible with uh honda link services. honda link remote services and released at the beginning of this year. this of course supports amity connection using the aws i mentioned earlier. and the major benefit is faster the motor operation, we are able to see 30 to 50% faster than the alco model. so it's a significant improvement. we think additionally, starting from these years are called ho link remote services is available for all three a go only the highest stream are supported. so that's so that every a code user can get the benefits such as so locking the car, even when you are leaving the parking lot or getting the comfort of a climate before you drive or to leave the by the mobile app temporary, even you if you don't have the key in the garage or the geo fence speed up as a parental control for your kids and security a can be detected by that.

so if you have dc as a code, you may want to consider joining handlin in conclusion, the benefit of the managed am using aws i is so who we managed service and the long live connection and the b scale level. and uh it's already in the operation in the production vehicle like off and easy to get start using data of console and legacy http asset can be utilized.

so this is a little bit old story. so we are already in the next stage. so i hope we can introduce and disclose that kind of new technology in same stage again. but my closing world is keep an open mind and for the new technology.

thank you, sir for joining me today.

thank you. yeah, thank you. so let's wrap up a few key takeaways uh for you before you leave the room. so number one, use aws iot services to enrich, modernize and add innovations to your connected vehicle platform. number two include aws in your journey. we love to collaborate and help builders build. and number three contact your aws account teams or us and schedule follow up meetings. we are super happy to engage with you.

and as laurie mentioned, there are a couple of related iot sessions here at re invent. during this week, you can see them here, you can take a picture uh memorize the numbers, uh join them. i'm sure you will enjoy them if you are an iot fan, that's all from us. and as laura said, we will be here uh for the next couple of minutes, join us.

um ah last but not least we are a data company. i keep forgetting. survey is super important for us. please take a minute or two to use your mobile app. provide a service. we would like to get your feedback so that we can deliver next next year and re invent uh an accession, a better one, fulfilling your needs, your expectations.

thank you very much and follow us on linkedin.

thank you very much. thank you."