00:00:00: [MUSIC]
00:00:07: >> Hello and welcome to the IAA Mobility Visionary Club.
00:00:10: I'm so excited to introduce our guest to you.
00:00:12: We are talking today with Dr. Karen Langana,
00:00:15: partner sales global director for
00:00:17: Automotive and Manufacturing at AWS.
00:00:21: She's going to be talking to us today from Michigan.
00:00:23: Great to have you with us, Karen.
00:00:25: >> Thank you, Sarah.
00:00:26: Thank you IA Mobility Visionary Club for
00:00:29: having me here today.
00:00:30: I'm very excited to be part of episode 10 to
00:00:34: talk about autonomous driving development.
00:00:37: We start with our industry,
00:00:40: our lovely automotive industry is facing a very rapid evolution.
00:00:45: This is being driven by software-defined vehicle,
00:00:49: a topic close to our hearts and you have
00:00:52: covered in previous episodes as well.
00:00:54: Driver assist technology leading into
00:00:57: autonomous driving and consolidation,
00:01:00: centralization of the electrical architectures.
00:01:04: The pace of innovation for our automotive companies,
00:01:08: it's really fast and they need to speed up their R&D or
00:01:14: research and development or product and engineering cycles.
00:01:18: This does require scalability.
00:01:21: It does require data-centric approach,
00:01:25: so they can stay competitive,
00:01:27: but most importantly,
00:01:29: they can meet the needs of the customers.
00:01:31: Customers today are seeking for innovation.
00:01:35: They see the car as an extension of our phones,
00:01:40: our automated houses, intelligent houses.
00:01:44: The challenges they face is because if they are legacy companies,
00:01:49: they do have legacy systems,
00:01:51: they see their data fragmented,
00:01:54: they see a siloed infrastructure,
00:01:57: that hinders innovation and the agility to
00:02:01: keep innovating on behalf of the customer.
00:02:04: There is a need for modernization,
00:02:07: and this is where digital transformation
00:02:10: combined with cloud migration can really help.
00:02:14: It's very critical to drive
00:02:16: the operational efficiency and speed and time to market.
00:02:21: In today's sections,
00:02:23: we are going to focus on how to modernize,
00:02:26: how to interject and bring AI,
00:02:29: artificial intelligence,
00:02:31: machine learning, generative AI,
00:02:34: data-driven insights to accelerate
00:02:37: autonomous driving development,
00:02:40: and create new business value.
00:02:42: This is the important part.
00:02:44: What do we observe in vehicle development?
00:02:47: It's a true convergence of hardware,
00:02:51: software, data, and artificial intelligence.
00:02:55: We see this in automotive industry,
00:02:58: which leads to what we call software-defined vehicles.
00:03:02: These vehicles, in fact,
00:03:04: they started their development completely on the cloud,
00:03:08: through a digital engineering,
00:03:11: digital twin, and allow us to do the process much earlier in terms of simulations,
00:03:17: in terms of defining the vehicle architecture,
00:03:21: in terms of defining the hardware that is needed.
00:03:24: That is a key shift from previously.
00:03:27: We know that the automotive industry operates on low margins,
00:03:32: so they used to decide what to have in the vehicle,
00:03:36: in terms of processing power,
00:03:39: storage power, based on cost.
00:03:42: You would really minimize what kinds of parts you have to reduce the cost.
00:03:48: Today, in a software-defined vehicle,
00:03:51: you try to decide based what is the buffer.
00:03:55: When you decide on storage and compute,
00:03:58: you want to make sure you leave space
00:04:02: to release new features through software
00:04:06: that will benefit the customer,
00:04:08: will make that vehicle more agile, safer,
00:04:12: more responsive to the customer needs,
00:04:15: and you do have memory and processing power available into the vehicle.
00:04:20: So, automakers, they really need,
00:04:23: in order to move towards a software-defined vehicle,
00:04:27: they really need to modernize their R&D functions.
00:04:31: By migrating to the cloud,
00:04:33: it enables rapid iteration,
00:04:36: it enables cost optimization,
00:04:38: it allows for scalable testing,
00:04:41: and leads to faster vehicle development.
00:04:45: In terms of AWS, we do offer that scalability.
00:04:49: If you think about AWS Elastic Compute Cloud,
00:04:53: our EC2 instances that run on the cloud,
00:04:57: they do offer a variety of options,
00:05:00: and some of them really mirror or mimic
00:05:04: the same hardware that you have in the vehicle.
00:05:07: For example, our Amazon EC2 DL2Q instances,
00:05:12: they are instances featuring Qualcomm AI100,
00:05:17: a hardware that is present in vehicle development.
00:05:21: You could think that we also offer a variety of other EC2 instances
00:05:26: that are powered by NVIDIA.
00:05:29: So, you are able to count on GPU power right on the cloud,
00:05:34: available in multiple regions to run your workloads.
00:05:38: The same as when we offer our EC2 instances
00:05:42: powered by Intel Exxon processors.
00:05:44: These are just some examples of the scalability of the cloud
00:05:49: and how AWS really partner with technology partners
00:05:54: to bring the important pieces for vehicle development
00:05:59: to our automakers.
00:06:00: Moving on a little bit, talking about security and compliance,
00:06:05: security is job zero for AWS.
00:06:08: So, AWS offers industry-leading security frameworks
00:06:14: to protect sensitive automotive data and intellectual properties.
00:06:19: So, that really gives a good base for facilitating global collaboration
00:06:26: and data integration.
00:06:27: So, by using AWS Data Lakes, you can unify the data from sensors,
00:06:34: autonomous systems, digital twins,
00:06:37: and you are going to improve predictive modeling.
00:06:40: You are going to enable continuous learning and autonomous systems.
00:06:45: On top of that infrastructure, we also see transformative technologies,
00:06:51: such as machine learning, artificial intelligence, generative AI.
00:06:56: So, in terms of services, you can count on Amazon Bedrock,
00:07:01: one of our services that will help you build, customize, and deploy
00:07:08: generative AI or GenAI models for rapid prototyping
00:07:13: and digital twin simulations.
00:07:15: You will be able to accelerate the design cycles
00:07:20: and you can optimize development costs
00:07:23: because you are simulating real-world scenarios
00:07:27: for virtually and basically anywhere where your development teams are located.
00:07:32: If you are thinking about machine learning, Amazon SageMaker,
00:07:38: it allows you to fine-tune GenAI models for predictive maintenance,
00:07:44: which is a big problem to be solved,
00:07:46: for accelerating vehicle design and for scenario analysis,
00:07:51: which is quite important when you are developing driver assist
00:07:55: or autonomous driving technologies.
00:07:58: So, by counting on those technologies, you will minimize the downtime
00:08:03: and you will optimize your research and development resources
00:08:08: by inserting the productivity feature into the product and into the development.
00:08:15: Amazon Q developer is a not-great example,
00:08:19: whereas GenAI accelerates software development
00:08:23: by bringing generative AI into the code development portion.
00:08:29: It will reduce the time and minimize errors in your code.
00:08:33: It will enable faster deployment of these complex autonomous functions.
00:08:39: And as I said, machine learning is also part of what we do at AWS.
00:08:45: You can count on Amazon SageMaker,
00:08:48: which develops and trains machine learning models
00:08:52: that can be applicable to object detection, sensor fusion,
00:08:57: and again, predictive analytics.
00:09:00: This enhances autonomous systems accuracy and reliability
00:09:06: and accelerates regulatory approval times, which are very important.
00:09:11: Automotive industry, per se, it is a highly regulated industry.
00:09:17: When we are talking about autonomous driving,
00:09:20: driver assist and safety, it's even more.
00:09:23: The standard is even higher.
00:09:26: If you're thinking about what is between the car or the cloud,
00:09:30: you can think about the edge.
00:09:32: So AWS IoT Greengrass Nucleus Light is one of the examples
00:09:38: that enables real-time data processing
00:09:42: and edge computing in connected vehicles.
00:09:45: And when we say connected vehicles,
00:09:48: they do have a connectivity tissue.
00:09:51: Usually through wireless connectivity,
00:09:53: it could be through satellite as well
00:09:55: that allows that connection vehicle to cloud.
00:09:59: All of this allows you to reduce the better the connectivity you provide,
00:10:05: the latency, and enhances decision-making.
00:10:09: If you observe a problem on vehicles that are on the road today,
00:10:14: you can immediately take action.
00:10:16: And if we're talking about a software-defined vehicle,
00:10:20: you can find the correction of the error,
00:10:23: and you can distribute that code through over-the-air updates.
00:10:27: So there is much more we can do.
00:10:30: Definitely by working with AWS or cloud infrastructure,
00:10:36: we can drive the future of autonomous driving.
00:10:39: We accelerate R&D transformation
00:10:42: by partnering with AWS and our 100,000 partners that we have.
00:10:48: You will be able to streamline your research and development.
00:10:53: You can unlock the power of artificial intelligence and machine learning
00:10:59: by using a selection of different types of services
00:11:04: that make sense for your architecture,
00:11:07: for the group of people that you have available in different regions.
00:11:12: And this will certainly allow you to explore several models,
00:11:17: to speed up testing and deployment.
00:11:20: And you can connect with AWS experts.
00:11:24: So we'll be present at IAA Mobility in September in Munich.
00:11:29: So I invite you all to stop by and see the demos from AWS service,
00:11:35: but also from our partners.
00:11:37: So for us, our partnership is really important
00:11:40: because they work with us on behalf of our customers.
00:11:45: So I invite you to explore AWS Automotive Solutions
00:11:49: and discover how we can enable you to transform your business
00:11:55: and accelerate autonomous mobility development.
00:11:59: And our future will be based on autonomous,
00:12:04: electric and software-defined vehicles.
00:12:07: Back to you, Sarah.
00:12:08: All right, Karen, thank you so much for giving us that State of the Union,
00:12:11: telling us a little bit about where we are and where we're headed.
00:12:14: I'm looking forward to seeing you in Munich in September.
00:12:17: I'm curious, Karen, how you see Generative AI transforming
00:12:21: automotive industry over the next, let's see, five years.
00:12:24: And I'm thinking not just about OEMs, but across the entire value chain.
00:12:28: Tell us where your thoughts are.
00:12:29: Very good that you emphasize or whoever submit the question
00:12:33: on the entire value chain,
00:12:35: because it is really not a problem and an opportunity
00:12:38: to not only for automakers, but for the entire value chain,
00:12:42: for our suppliers, for the dealers that operate.
00:12:46: And where Generative AI can really help,
00:12:49: I will mention a few examples, such as supply chain optimization.
00:12:54: And I'm starting with supply chain because in that domain,
00:12:58: we are seeing a lot of instability given the care of exposure
00:13:02: to our industry, our suppliers.
00:13:05: So Generative AI can really help in making fast decisions
00:13:10: and you may need to really switch your sourcing
00:13:14: based on the risks that you are trying to minimize.
00:13:18: You can also help predict supply chain disruptions
00:13:23: when you are using data in a Gen AI.
00:13:26: As I said, you can optimize your sourcing strategies
00:13:29: even to try to meet your sustainability target goals.
00:13:34: The other area where we see Gen AI, it is on customer experience.
00:13:38: So people, they want the vehicle to have their unique
00:13:43: characteristics or to be an extension of the settings they have at their home.
00:13:49: So the personalization, the recommendation, the alerts, the instructions to better drive
00:13:56: your car or understand the problems of your car is another area.
00:14:00: So in the customer experience, GenAI can really help transform.
00:14:06: As I mentioned in my introduction in product design in engineering, AI-driven simulations
00:14:12: can really reduce prototype cycles, can enable faster development, reduce dependencies that
00:14:20: you have on hardware availability because you can rely more on digital twins.
00:14:26: And also in the production space in terms of smart manufacturing and quality control,
00:14:32: by bringing GenAI to the edge to the production line, you can detect defects very early in the
00:14:40: production line, avoid those to be propagated to other products.
00:14:44: You can also work with maintenance alerts for your machinery on the floor.
00:14:51: So these are some of the areas where the list could be much longer, but I think these are
00:14:56: the more relevant ones.
00:14:58: That makes sense.
00:14:59: Karen, I'm wondering if you would share with us some of the more exciting use cases that you
00:15:03: have personally seen where AWS partners have enabled innovation in either electric or
00:15:09: autonomous vehicles.
00:15:10: So one of them is, as you said, an autonomous vehicle or driver assist technology.
00:15:17: So we are seeing in more recent years a lot of emphasis on driver assist towards higher
00:15:23: level of autonomy.
00:15:25: We still have companies solely dedicated on autonomous driving, but we're seeing a lot
00:15:31: of increasing the driver assist in that case.
00:15:34: So our collaboration with startups and even with our automakers in implementing AI models
00:15:42: for real time object detection for agentic AI, Cinecreations is one of the areas.
00:15:50: And if you Google, you will be able to see some of that.
00:15:53: So AWS does invest in strategic collaboration agreements with companies like that.
00:16:00: Another area that is very important, as I said at the beginning, in the software defined vehicle,
00:16:06: a virtual engineering workbench where you can have like hardware in the loop testing
00:16:13: that enables faster autonomous model deployment and development.
00:16:18: It's really a cool area because you really virtualize a lot of the SDKs that we used to
00:16:26: have pieces of hardware on top of the engineering desks.
00:16:30: They become a part of your account on the cloud infrastructure.
00:16:35: The other one, because of this trend to move into electric, the battery management systems
00:16:42: also relies and there is a lot of predictive analysis and also learning the charging habits
00:16:49: of the customers, combined on how you can extend the lifetime of your battery.
00:16:55: Those insights are really, and we do have partnerships there.
00:16:59: We also use our in-house knowledge.
00:17:03: As you can imagine, our Amazon e-commerce has a very large fleet of vans for last mile delivery,
00:17:10: middle mile, and we do learn from operating those electric fleets.
00:17:16: So that analytics can also be passed to our customers.
00:17:20: I want to talk a little bit about product development with you, Karen.
00:17:23: And I'm curious if you could share with us how you see the convergence of hardware and
00:17:28: software development evolving over the next, let's say, five to 10 years.
00:17:32: Where are we headed?
00:17:33: So interesting question because we do see hardware, software, AI, and analytics
00:17:41: converging in terms of the contextualization, I would say.
00:17:48: The hardware within the context that is part of, the software within the context that is part of.
00:17:56: So these are very important to understand where you are using both.
00:18:01: And they do converge in the sense that you can have the hardware today on the cloud
00:18:08: that will be part of your product later, either in a car or even on a smart appliance,
00:18:15: for example. You could have your smart refrigerator at home where you also see that conversion of
00:18:22: you are developing or a hardware specific for a context, and you are developing the
00:18:29: software that runs on that same hardware. So you see a lot of development on the cloud,
00:18:36: accelerating the cycles and bringing them together in terms of the context that they are part of,
00:18:43: in terms of testing the integration of both, which is really important.
00:18:48: But you have the benefit of virtualization or digital twins that allow you to test that
00:18:56: before the real world.
00:18:58: No, Karen, I understand that you have a vision for how AWS's partner ecosystem can accelerate
00:19:04: innovation beyond what any single OEM or supplier could achieve alone.
00:19:10: Can you share that vision with us?
00:19:12: So one, as I mentioned a few times, is we do develop services that are industry specific,
00:19:18: so that will work on that or were developed for the industry. And in this case, it could
00:19:24: automotive, but we work with other industries. So we learn from other industries, and there is
00:19:31: that cross industry knowledge that is passed on. So that is one. The second is really the
00:19:37: richness of our AWS partner network, where we have technology partners, industry partners,
00:19:45: and we partner with them through strategic collaboration agreements, which are not only
00:19:52: monetary investments, but is really an investment of knowledge and infrastructure to either enhance
00:19:59: if they are already a partner present in the ecosystem. We have many of those that didn't
00:20:06: run their tools for research and development on the cloud, and now they run on the cloud.
00:20:11: For example, seniors, vector, G space, examples of partners where they do offer their tools.
00:20:19: We also collaborate with global system integrators who understand our partner business. So they are
00:20:26: also part of our SCAs, a collaboration agreement in terms of CAP Gemini, Deloitte,
00:20:35: Weepro, are examples of where they are also developing intellectual property to serve the
00:20:45: industry from the business lenses, from the solutions lenses. So that enriches the ecosystem.
00:20:53: So our customers, if they are truly, truly builders, then can rely on AWS services. If they have
00:21:01: already their partners of preferences, they're most likely to be already our partners. If they
00:21:07: already use these tools that have been present in the industry for many years or decades, we will
00:21:14: help them with the different same tool, but now running on the cloud. So enriches the options,
00:21:21: optionality to our customers, which is one of the basis of the AWS and Amazon, gives the customer
00:21:28: options what best fits their needs. Given your own experience, Karen, at an OEM,
00:21:34: what are some of the most valuable insights that traditional OEMs have gained as they embrace
00:21:39: digital transformation? So one of them is to be able to become more agile, either through
00:21:47: software development, or as I said, even changing, like when they adopt virtualization, the digital
00:21:54: workbench. So they can become faster in accelerating their development cycle. The other one is the
00:22:04: culture of innovation, which really fosters that you try things or fail fast, as we tend to say.
00:22:12: So you have a different way to assess risk. I'm not saying fail in any case, regardless of the
00:22:19: risk. It is how you assess and approach risk in terms of developing faster prototype, or I said
00:22:27: relying more on simulation, very close, accurate simulations to make decisions. So agility is one,
00:22:38: the culture of innovation. And the third one, very important, it shouldn't be the third one,
00:22:43: is listen to the voice of your customers. Because before the relationship with the customer was
00:22:50: mostly through dealers, and we see the modern OEMs, Chinese OEMs selling directly to the customer,
00:22:58: and keeping that relationship through the life cycle of the product. So the digital transformation
00:23:05: brings that customer data, or customer 360, to the front of the vehicle development. So you're
00:23:13: listening to the customer signals and developing solutions that really meet their needs. So you
00:23:21: may offer a cooling system in Florida, but you wouldn't offer maybe a cooling system,
00:23:27: wouldn't be a number one feature if you live in Michigan, or in another code location.
00:23:33: That's very good illustration. Different customers, different needs.
00:23:37: What areas in autonomous systems research do you think still require fundamental breakthroughs
00:23:44: before this true level four, level five adoption can scale? So I still believe, I think Eliather is
00:23:51: one of the technologies that really improves the accuracy, decision making, lots of factors in
00:23:58: autonomous driving. There are other professionals that believe you can do everything with cameras,
00:24:04: so they are all in development stage, but one of them is Liather. Solid state Liather is a technology
00:24:12: that is still, it does not have any moving parts, so it makes it more reliable and less susceptible
00:24:21: to failure. So that's where we still need some work. We need to see more advanced of the
00:24:27: technology in terms of hardware, software availability, production cost. It is one.
00:24:33: The other one is in terms of realistic 3D environments, so high fidelity models,
00:24:40: scenario generation, increase the focus on safety. Because if you're, even if you're talking about
00:24:48: autonomous driving based on trucks, it is easier on a highway if you're thinking on an urban scenario,
00:24:56: where you can have a bicyclist coming behind a van that is parked on a bus stop, which is not
00:25:03: supposed to be parked, but making a quick delivery. So the complexity of the urban scenarios and being
00:25:11: able to have this realistic 3D environments or platforms that you can test and retest
00:25:19: and extract things that are more important to validate safety. I think these are two of the
00:25:25: areas where I would say we still need to see more breakthroughs. That makes a lot of sense.
00:25:30: Karen, listening to you today, I'm struck by how many times we've used the word listen and also
00:25:36: collaboration. And so as we come to the end of our session, I want to ask you as a last question,
00:25:41: how is AWS helping to standardize platforms, protocols, data models to better enable collaboration
00:25:49: across the entire value chain? So one is we do participate in organizations or
00:25:57: groups that are defining standards. So Eclipse is one of them. Eclipse Foundation is one of them,
00:26:06: where AWS is part of. Katina X is another one. So Eclipse Foundation, very key for software
00:26:13: defined vehicle development. So helping to be together with experts and potentially influence
00:26:21: standards. Katina X for supply chain for battery management, as I mentioned, so AWS is a member
00:26:29: since 2023. Again, by being together with our customers and the players and others that set
00:26:36: industry standards allows us to have a seat in that environment. We also participate in the NADA
00:26:44: for dealership standards as well. In all these examples, we have more where we are present
00:26:51: to learn from our customers and suppliers, but also to try to drive standards that will
00:26:57: accelerate development. All right, Dr. Karen Langana, thank you so much for being our guest today
00:27:03: on the IAA Mobility Visionary Club. It was so great to chat with you about collaboration
00:27:08: and listening. And I hope to see you in Munich in September at the big event. Thanks so much.
00:27:14: We are back very soon next week with episode 11. I'll see you then. Take care.