Recap of AWS re:Invent 2022 for the Automotive Industry

At AWS re:Invent 2022, which was held in Las Vegas, Nevada, from November 28 to December 2, AWS made several announcements relevant to the automotive industry. This blog post will summarize all the services and features from these announcements that are most relevant to the automotive industry. We will highlight AWS automotive and manufacturing industry business unit (IBU) team’s breakout sessions, demos, workshops and chalk talks that can benefit auto industry participants. This will help you overcome key challenges as the automotive industry continues to reinvent itself to offer more connected, sustainable, personalized and overall, a better mobility experience for customers.

In Automotive Reinvention (AUT101-L), Wendy Bauer (General Manager, AWS Automotive) highlighted the macro trends that are driving the automotive industry, which are as follows.

• Sustainable mobility: Customers around the world expect and demand that the automotive industry should provide sustainable mobility for years to come around the world.
• Technology in the Vehicle: Connected, Autonomous, Shared, Electrification (CASE) is transforming the industry like never before. Today, approximately 246 million connected cars are on the road globally. Connected mobility relies on the power of data and on edge computing, where cars can make decisions in real time. Customers can interact with their vehicles and the vehicles can communicate with their environment, including other vehicles, pedestrians and infrastructure.
• Customer experience and personalization: Consumer needs have evolved from being price-driven in the past, to valuing experience and personalization. Consumers are on the lookout for the best consumer experience per mile having infotainment, geo-localized and personalized content, and a seamless experience.

In the same session, Marc Solsona Palomar (General Manager, Automotive Technology, AWS) announced the availability of Blackberry QNX on AWS graviton, which enables cloud-native development of embedded software and helps reduce hardware dependency for the development teams in automotive software development. In addition, the AWS Industry Products and Blackberry teams showcased a Blackberry IVY demo at industry demo area. Deploying IVY onto the latest Bosch infotainment electronic control unit (ECU), AWS and Blackberry team developed fueling and predictive maintenance demo along with Blackberry IVY partners. IVY collects vehicle sensors data, performs edge computing and sends insights to AWS cloud.

The Automotive and Manufacturing industry business unit (IBU) within AWS works with several major Original Equipment Manufacturers (OEM), Tier1’s, customers and partners in the automotive space and focuses on several strategic workloads. We will provide further highlights and announcements from re:Invent 2022 with respect to some of these strategic workloads.

Autonomous Driving

Dr. Werner Vogels, Vice President and Chief Technology Officer at Amazon, in his keynote encourages us to build asynchronous, loosely coupled, event-driven systems that have controlled concurrency and controlled parallelism. This would enable us to build evolvable architectures. Dr. Vogels states “With autonomous driving, simple 2D maps are not enough”. Vehicles need to have spatial intelligence and it requires a fusion of sensors and custom ML models and the perception system has data that comes from cameras, LiDAR and infrared radar that create a 360-degree field around it. AWS Ambit scenario designer which is an Open-source suite of tools to streamline 3D content creation at scale is available on github. It can render near-real-time 3D models and environments for use in simulations, training, and virtual walkthroughs. With an example of Aurora, who is building technology for autonomous trucking and use LiDAR, Radar and cameras and performed over 2 million unprotected left hand turns virtually before they did one in the real world, Dr. Vogels encouraged industry participants to think about spatial simulation to solve real-world problems. While simulations require a significant amount of memory, AWS SimSpace Weaver allows customers to run massive spatial simulations without managing infrastructure. A user could simulate traffic patterns or public transportation networks or supply chain infrastructure while SimSpace Weaver handles all the networking, the memory management, and synchronization.

Autonomous vehicles rely on geospatial data to navigate from A to B. Frequent updates on construction zones and changes in laws and driving regulations are enabled by maps powered by geospatial data. However, geospatial data is very difficult to work with. Accessing high-quality datasets to train ML models is challenging and requires working with multiple data sources and multiple vendors, time consuming preparation of massive datasets and having limited visualization tools. To unlock the value of geospatial data, in his keynote, Dr. Swami Sivasubramanian, VP, Data and Machine Learning at AWS, announced that Amazon SageMaker now supports new Geospatial ML capabilities making it easier to build, train and deploy machine learning models using geospatial data. With this capability, customers can acquire geospatial data from readily available geospatial data sources, including satellite imagery, maps, and location data with just a few clicks, and can easily prepare geospatial data with built-in algorithms. Amazon Sagemaker also provides built-in pre-trained neural nets to accelerate model development. Giving an example of how to predict dangerous road conditions due to rising water levels, Kumar Chellapilla, GM, ML/AI services at AWS demonstrated how to make this prediction using geospatial capabilities of Amazon Sagemaker.

In AI/ML session (AIM217-L), Bratin Saha, VP, ML and AI services, AWS, explained how the automotive industry uses geospatial data. Invited to the talk was Marco Görgmaier, GM, Data Transformation and AI at BMW, who spoke about how BMW group is using Amazon SageMaker geospatial machine learning capabilities (such as map matching, efficient geo hashing or digital maps) to build a sustainable solution to predict the affinity of drivers for an electric vehicle based on their driving behavior with an accuracy of more than 80%. Having gone from idea to solution in just 8 weeks, he talks about the advantages of having a single API to access, transform and enrich disparate geospatial data and having pre-built visualizing tools specifically tailored for geospatial data.

In the industry demo space, the AWS professional services team show cased the Autonomous Driving Data Framework(ADDF) which is a collection of modules for scene detection, simulation (mock), visualization, compute, storage, and centralized logging deployed using the SeedFarmer orchestration tool. ADDF allows customers to build distinct, stand-alone Infrastructure as code modules and exchange information about dependencies using metadata which can be exported from one module and imported into another. AWS DeepRacer League is the world’s first global autonomous racing league, driven by machine learning (ML). Developers of all skill levels can compete in races online and in person worldwide. Over 150,000 developers participated in AWS DeepRacer in 2022 from over 160 countries around the globe. AWS DeepRacer 2022 Championship Speedway race was held at the re:Invent 2022.

High performance compute

One of the primary use cases where automotive manufacturers use high performance computing is for simulation workloads for autonomous driving and driver assistance systems. This year, Peter DeSantis, SVP, AWS Utility Computing in his keynote made several announcements of EC2 instances for highest compute performance such as Amazon EC2 R7iz instances for running X86 based workloads, EC2 instance C7gn with enhanced networking and powered by Graviton and the Nitro v5 chip. HPC workloads tend to be compute-, network- and data- and memory- intensive. Dave Brown, VP, Amazon EC2 in the Compute session (CMP223-L) announced Amazon EC2 HPC6a instances for compute-intensive HPC workloads, Amazon EC2 HPC7g instances powered by Graviton3E and Elastic fabric adapter for compute- and network- intensive HPC workloads, and EC2 HPC6id instances for memory and data-intensive HPC workloads. Also announced were EC2 P4de instances for ML training and HPC applications in the cloud, network optimized tranium based EC2 Trn1n instances for training ultra-large models, EC2 Inf1 instances powered by AWS Inferentia chip for machine learning inference for small to medium complexity models, EC2 Inf2 instances powered by the Inferentia2 accelerator for ML inference of large complexity models.

Easy Development

With the shift from a mechanical to a software mindset, automotive systems today have more software than ever before. To simplify and accelerate software development, Dr. Werner Vogels announced Amazon CodeCatalyst (preview), which takes away the heavy lifting that sits around development. Amazon CodeCatalyst is a unified software development service that makes it faster to build and deliver on AWS. With the basic concept of blueprints which can be individual or organizational, it not only creates code but also sets up everything from project files, configuration of integrated tools, source code control, CI/CD pipelines, issue management and integrating with GitHub. It enables customers to create a project with everything they need in minutes. It also makes it easy to switch between environments and codebases with one click. Another feature is the AWS application composer (preview) which simplifies and accelerates architecting, configuring and building serverless applications. It allows customers to visually design and build serverless applications quickly. It also enables our customers to maintain a model of their architecture that’s easy to share and build with team members, clients or colleagues.

Addressing the need to build bigger applications with smaller components while using a standard interface, Dr. Werner Vogels announced Amazon EventBridge Pipes which connects event producers and consumers in seconds, and allows users to stitch AWS services together, and is designed for integrating messages from different AWS services. Amazon EventBridge Pipes can manipulate messages that flow through the pipeline before they reach the consumer with an AWS Lambda or an AWS Step Function. It also has built-in filtering to send only a subset of events to flow to consumer.

To simplify machine learning development, Bratin Saha in his AI/ML session (AIM217-L) announced the launch of the next generation of SageMaker Studio Notebooks which makes it easy for customers to visually prepare their data, to do real-time collaboration (develop, co-edit and collaborate in the notebook itself), and to quickly move from experimentation to production with automated notebook deployment.

Vehicle toolchains

In Automotive development toolchains (AUT203), Tara Vatcher, SVP, Software Architecture & Development – Platform Stellantis and Hendrik Shoeneberg, Principal Data Architect at AWS talked about the collaboration between Stellantis and AWS to develop the Virtual Engineering Workbench(VEW). VEW enables customers to transform automotive software development from the physical world to the virtual world. It ensures that all developers from different geographies have access to a consistent environment with the right tools and provides a virtual environment with fully automated pipelines, reducing hardware dependency and enabling the ability to shift left on the V-model.

Customer testimonials

During the Autonomous driving simulation and validation on AWS (AUT204) session, customers from Volkswagen Commercial Vehicles talked about how data-driven development solution for self-driving systems, ‘SNOWPARK’, built on AWS, has helped democratize data and make it accessible to everyone in the brand. In the same session, customers from Toyota Motors North America explained about how their scale-out edge-to-cloud DataOps solution for ADAS/AV programs in production built on AWS helped improve development processes, shortening time from ingest to analysis from months to days.

Building connected vehicle and mobility platforms with AWS(IOT311) explained about how AWS IoT Core enables large-scale connected vehicle workloads, discussed best practices for Connected Vehicle Platforms and described how AWS is helping OEMs, tier 1 suppliers, fleet telematics solution providers, and automotive ISVs build and deploy systems that securely connect vehicle fleets to the cloud. The session also included customers from Mercedes-Benz sharing their experience connecting a large fleet of vehicles to AWS IoT Core.

Automotive manufacturing

Dr. Werner Vogels also announced AWS Step Functions Distributed Map in his keynote, which orchestrates large-scale parallel workloads within serverless applications. With the AWS Step Functions in Distributed Map, users can do a map and reduce step very easily and process huge data sets of semi-structured data by using simple lambda functions. For example, in automotive manufacturing, where machines and robots on the production shop floor generate thousands of files per second, it is easy to process and deduce simple outputs such as average torque pressure by processing all the files at once using AWS Step Functions Distributed map.

Digital twins give customers insights into all aspects of their automotive production line and manufacturing process. AWS IoT TwinMaker makes it faster and easier to create digital twins of real-world systems and apply them to monitor and optimize industrial operations.

In AI/ML session (AIM217-L), Bratin Saha, VP, ML and AI services, AWS, spoke about several ML-powered use cases, one of which is predictive maintenance of industrial equipment to significantly reduce equipment downtime. Amazon Monitron, launched in 2020, comes with wireless sensors that measure machine’s vibrations and temperature, a gateway and an app, and uses ML to predict when a piece of equipment needs maintenance. Best of all, it provides all of these capabilities without the need to write a single line of ML code.

Analytics

Just last year, Amazon QuickSight which is an ML-powered business intelligence (BI) service launched 80 new features. This year at re:Invent, AWS announced print-friendly highly-formatted reports in QuickSight that help customers create operational paginated reports. This is especially useful in automotive manufacturing and process planning, where, workers in the factory are given printed instructions for example on how a specific model of a car needs to be assembled on the shop floor. Amazon QuickSight makes it easy to generate and print hundreds of paginated report pages that can be handed over to the shop-floor workforce.

Another popular feature announced in 2020 is Quicksight Q, which gives anyone in an organization the ability to ask business questions using natural language. For example, you can ask Quicksight Q, “What are the top 10 products by sales in 2022” and Q will return the answer in the form of visualizations in seconds instead of the developer creating dashboards and writing SQL queries. This year, AWS announced ML-powered forecasting with Q to which allows customers to ask Q to forecast and can also ask “Why” questions with Q so they can dig deep and understand past events and analyze trends that have impacted the forecast.

In his keynote, Dr. Swami Sivasubramanian emphasized on the importance of maintaining data quality in data lakes and announced the preview of AWS Glue Data Quality using which enables customers to generate automatic data quality rules and to measure, monitor and manage data quality in their data lake. Today, many customers use Apache Spark, an open-source framework for complex data analysis like regression testing, or time series forecasting and for building distributed applications. However, it requires managing infrastructure to run interactive analytics. To simplify building interactive applications, rather than dealing with infrastructure setup and keeping up clusters for interactive analytics, Dr. Swami announced Amazon Athena for Apache Spark which allows users to start running interactive analytics on Apache Spark in under one second. It enables customers to spin up spark workloads up to 75 times faster than other serverless Spark offerings. To make data sharing easier, Dr. Swami announced Amazon DataZone, a data management service that helps customers catalog, discover, share and govern data across the organization, not just in AWS but also data residing in third-party data services all while meeting their organization’s security and data privacy requirements.

Zero ETL Vision

One of the places where customers spend the most time is in building and managing ETL pipelines between transactional databases and data warehouses.
Adam Selipsky, CEO of AWS, in his keynote, announced fully-managed new zero-ETL integration between Amazon Aurora and Amazon Redshift, which enables near real-time analytics and ML on transactional data. With just a few clicks, data is seamlessly made available in redshift within seconds. Customers can replicate data from multiple Aurora databases into a Redshift instance, and data is automatically and continuously updated in near real-time. The entire system is serverless and dynamically scales up and down. Another announcement came with the integration of Amazon Redshift with Apache Spark which enables customers to easily run Apache Spark queries on Redshift data from Amazon EMR, AWS Glue, and Amazon SageMaker, without the need to move any data or build any ETL pipelines. Both of these announcements solve major ETL pain points for our customers and enable us to take two steps forward towards AWS’s vision of a zero ETL future.

In-Vehicle Customer experience

In the automotive industry, one critical element of customer experience is when it comes to roadside assistance, or emergency response. Now, a customer’s experience with a contact center can be automated using Amazon Connect. This year, AWS launched three new capabilities of Amazon Connect namely ML driven forecasting, capacity planning and scheduling. These new capabilities help contact center managers to optimize agent schedules and ensure that they have the right agents at the right time. Second, AWS is previewing agent performance management capabilities with Contact Lens and real-time call analytics. These tools help customers reduces the time that contact center managers spend identifying performance issues and help in coaching agents. Amazon Connect now includes a new user interface that guides agents through customer interaction with step-by-step actions, helping agents resolve issues even faster. Another new feature Contact Lens for Amazon Connect provides is conversational analytics that can be used to address customer issues, define, assess and improve agent performance, and identifying crucial company and product feedback.

The Digital customer engagement for automotive (AUT201) session included customers from Rivian who spoke about using Amazon Connect to build their Customer Engagement Center, Rivian Guide, Rivian Vehicle Services, designed to enable an elevated customer experience on communication channels with the customer. This session also included customers from BMW who spoke about achieving customer centricity from a data perspective by building Cloud Data Hub (CDH) on AWS to curate personal experiences for their customers. To improve digital customer experience continuously, OEMs need to update vehicle software regularly to update in-vehicle applications and add more functions in the vehicle. AWS partner Harman showcased Over-The-Air(OTA) update demo at the industry demo area using AWS services.

Supply Chain

In his keynote, Adam Selipsky talked about supply chains as coordinating complex global networks of suppliers, parts, manufacturing sites, distribution facilities and transportation providers all while trying to deliver the right goods at the right time at the lowest cost. Managing inventory requires that customers have accurate and up-to-date visibility into their supply chain. To get a complete view of inventory and of their supply chain, customers need to build custom integrations to collect and process data across a vast array of ERP’s and supply chain management systems. To meet these needs, Adam announced the AWS Supply Chain (preview) that is designed to improve supply chain visibility and provides actionable insights to help customers mitigate supply chain risks and lower costs. With this service, customers can get a unified view of their supply chain data, ML-powered insights, recommended actions, and built-in collaboration capabilities so they can react quickly to unexpected issues. With just few clicks, they can connect to their supply chain data from SAP S/4 Hana or EDI or SAP ECC, and AWS Supply Chain automatically sets up a data lake using ML models that have been pre-trained to understand, extract, and transform disparate, incompatible data into a unified data model. AWS Supply Chain then contextualizes the customer’s data into a real-time visual map, and this map shows overall inventory health across the customer’s supply chain network. Customers can also create a watchlist to get notifications when inventory may be at risk. AWS Supply Chain helps customers mitigate risk and lower costs by giving them a unified view of their supply chain and surfacing the best actionable insights with pay-as-you-go pricing and no upfront licensing fees.

Conclusion

This blog post brings you AWS announcements that are most relevant for the automotive industry. We encourage you to review your workloads and find out which of these announcements can be adopted to your workloads. If you want to explore how these new products and features can enhance your organization’s agility and efficiency, AWS is here to help.
Learn more about our offerings at the AWS for automotive page, or contact your AWS team today.

Related resources:
AWS cloud solutions health industry