In our prior blog, we discussed a definition and framework for Digital Twins consistent with how our customers are using Digital Twins in their applications. In particular, we described a Digital Twin leveling index to help customers understand their use-cases and the technologies needed to achieve the business value they are seeking. This Digital Twin leveling index is analogous to what we see in the self-driving cars space which uses an L0 through L5 system, where L0 is manual driving, L1 is cruise control, and L5 is a true autonomous self-driving car with no steering wheel.
As a recap, the figure below shows an overview of the four Digital Twin levels which we described in the prior blog. In this blog, we will illustrate the L1 Descriptive level by walking through an example of an electric vehicle (EV). You will learn, through the example use-cases, about the data, models, technologies, AWS services, and business processes needed to create and support an L1 Descriptive Digital Twin solution.
L1 Descriptive Digital Twin
An L1 Digital Twin focuses on describing the physical system and includes content ranging from the initial business case analysis and product requirements to the detailed engineering design and 3D visualizations. The intent of an L1 Digital Twin is to describe the structure so that the user can understand the components and the relationship between those components that make up the physical system. In many cases, for L1 Digital Twins, the physical system doesn’t exist yet as it is still during the early design phase. Alternatively, the physical system might exist, and the L1 Digital Twin is used to understand the average behavior of the physical system under a specific set of operating conditions (such as in a computational fluid dynamic or solid mechanics analysis). Examples of L1 Descriptive Digital Twins that we’ll describe below include 1/ asset models, 2/ engineering design, and 3/ immersive virtual reality environments.
When considering a physical entity (or an asset), the most basic description consists of identifying all the sub-assemblies and components that make-up the system. For example, in the diagram below, we show a hierarchical representation of an EV vehicle. The diagram shows the major systems, subsystems and components that make up the vehicle. Note the diagram below is for illustrative purposes only and is not exhaustive. This information describes the structure of the Digital Twin and is usually represented as an entity model in a graph database such as Amazon Neptune, and can be queried and aggregated with data from other sources using a knowledge graph. Note that many industries have developed standards to enable interoperability between different systems and across the life cycle of the asset. For example, ISO 19650 and UK 1192 are international standards developed for building information modeling. In practice, keeping the configuration details updated is a common challenge for equipment operators and knowing the configuration of an asset is key for use-cases such as maintenance scheduling, inventory planning, and issuing product recalls for safety reasons. AWS IoT TwinMaker is a service that makes it easier for developers to create digital twins of real-world systems such as buildings, factories, industrial equipment, and production lines. In particular, it makes it easy to connect to different data sources to create, update, and query the asset entity model. This data can then be displayed in a visualization or be routed as input data to a machine learning or physics-based model to make predictions.
From an engineering design perspective, there are many automobile subsystems that need to be designed such as the structural frame, the power train, the electrical system, the steering and suspension, as well as the exterior (both for styling and aerodynamics), and the interior (for styling and comfort). Each of the subsystems are engineered using a variety of computational tools that have been validated over decades of experience. For example, the overall ride handling and vehicle dynamics is modeled using a physics-based multi-body dynamics simulation that models the stiffness of each of the structural components of the assembled vehicle.
In the example below, provided by AWS Partner Maplesoft, the tire-road contact interaction is modeled as the vehicle is driven over simulated uneven terrain. The inputs to the simulation include the vehicle speed and steering inputs, and the model provides the deflection of the vehicle suspension. The simulation takes into account the details of the tire stiffness and the design of the suspension to understand when the tire-road contact forces are too low to provide adequate traction. The vehicle dynamics and stability during maneuvers (such as braking on slippery roads) for an EV are quite different than for an internal-combustion engine for many reasons, including the bottom-heavy weight distribution of an EV (since the battery packs are under the floor) versus conventional vehicles where the engine is in the front. Dynamic system simulations enable the engineer to simulate the vehicle dynamics under different conditions to meet the engineering requirements for the design. In this case, we also provided a 3D rendering of the vehicle for enhanced visualization.
For EVs in particular, performance analysis of the electrical drivetrain is critical in order to make sure the vehicle meets customer’s expectations in regards to charging times, acceleration, and range. For example, the battery range of an EV on a cold day driving up and down hills is very different than the range on a warm day on flat roads. To explore the EV powertrain performance, we modified the previous EV model as shown in the figure below. In this model, we included the relevant parts of the driveline and a simplified vehicle model for aerodynamics and terrain inclination. The graphs in the image show the key battery and motor parameters (voltage and current) as the vehicle follows the simulated drive cycle, and the battery state of charge (SOC) can be shown to decline as the battery discharges. This type of analysis enables the design engineer to understand if the electrical drivetrain has the right performance characteristics to meet or exceed the expected route profiles for the target customers. To interact with a variation of this model, please visit the Maplesoft website.
Other examples of engineering analysis using different software tools include computational fluid dynamics of airflow over the vehicle to minimize aerodynamic drag, finite element analysis of the structural components to make sure they have the strength to endure the vehicle loads, and ride-comfort analysis to simulate the passenger vibration/comfort during vehicle operation. AWS Partners include a broad range of Independent Software Vendors (ISVs) that provide engineering simulation software for computationally intensive workloads such as computational fluid dynamics (CFD), finite element analysis (FEA), drug discovery, weather modeling, electronic design automation (EDA), and others. These ISV solutions can be deployed using AWS High Performance Computing (HPC) services such as AWS ParallelCluster or through AWS Marketplace. In addition, AWS HPC Competency Partners also provide a fully managed cloud HPC environment, and end-to-end cluster provisioning, deployment, management, and support for customers to run HPC workloads on AWS. Using simulations on AWS HPC infrastructure lets manufacturers reduce costs by replacing expensive development of physical hardware with virtual models during product development. More generally around the topic of engineering design, we’re hearing a lot from our Original Equipment Manufacturers (OEM) customers about the desire to modernize their engineering workflows in their companies and are looking to AWS to provide guidance on digital transformation. AWS is investing in Digital Engineering efforts to address key pain points related to data and model-sharing across diverse engineering tools, across internal functional groups, and across the external supply chain with robust permissions management.
Immersive Extended Reality Experience
The L1 Descriptive example above focused on the engineering description during the design of the vehicle by the OEM. Another compelling L1 use-case is the application of extended reality (XR) technologies, such as high fidelity 3D, immersive virtual reality (VR), and interactive augmented reality (AR), to immerse users into a true to life vehicle experience during the customer engagement process. In this example, AWS Partners Cavrnus, Theia Interactive, and Epic Games have developed a metaverse solution to create an immersive collaborative replica of a showroom featuring an Audi© A5 convertible where brand representatives can create more meaningful interactions with their customers through an always-on and available virtual space.
For example, a vision for the future of car buying is to have the customer enter an immersive virtual showroom from the comfort of their home, and have a brand representative join in the same virtual showroom to guide them through the sales cycle in a metaverse experience. To highlight this, we’ve taken screenshots from the Cavrnus immersive experience. We can see the customer viewpoint as they walk down the hallway to approach the vehicle, then see the vehicle in the showroom, and interact by opening the door and sitting inside the vehicle.
The Cavrnus experience enables multi-user collaborative environments where all users have full spatial 3D co-presence including voice and video streaming. The customer can invite their friends and family to join the immersive experience remotely from their own locations and interact together through avatars, voice and video. The customer is able to change the color of the vehicle, the trim level, wheels, and the interior per the configuration options available. All users will see the changes immediately, and can engage in collaborative conversations and use whiteboards to explain their ideas to each other. They can then render out their own personal commercial for their exactly configured vehicle.
In the images below, the customer continues to interact with the vehicle by opening the convertible roof and the trunk. The customer also modifies the vehicle color and interior trim options while engaged in collaborative discussions with their friends and family who can be seen via video and avatars..
This experience is fully interactive and the customer is able to use their controls to “walk” around the virtual showroom as if they were there in-person. For a video showing the immersive experience, please visit the Cavrnus website and see for yourself!
For this solution, the digital showroom environment was created by Theia Interactive. The showroom assets were built by their digital artists using digital content creation tools. The vehicle and vehicle components originated from the original CAD and engineering data provided by Audi©, and processed through a conversion/optimization workflow. The experience is powered by Cavrnus, and hosted on AWS infrastructure including the use of G4 and G4ad EC2 instances using Unreal Engine for rendering and streaming. The experience can also be pixel streamed from AWS GPU servers in the cloud to any end device via a web browser, including smart phones.
AWS is continuing to work with customers across industries to simplify the process of 3D content generation, storage, hosting, and experiences. For example, Amazon Nimble Studio is a virtual studio, now available for digital content creators, providing virtual workstations, shared storage, and a built-in rendering solution that connects creative tools in media and entertainment production environments. Additionally, AWS in partnership with Epic Games, has recently begun offering an Unreal Engine Amazon Machine Image (AMI) available in the AWS Marketplace that makes the process of developing Unreal Engine based content turnkey.
We, at AWS, are very excited to support our customers in their Digital Twin journey with a range of AWS services and ISV partnerships to support use cases across all four Digital Twin levels. In this blog we described the L1 Descriptive level by walking through an automotive example. In future blogs, we will extend the automotive example to demonstrate L2 Informative, L3 Predictive and L4 Living Digital Twins.
About the author
Dr. Adam Rasheed is the Head of Autonomous Computing at AWS, where he is developing new markets for HPC-ML workflows for autonomous systems. He has 25+ years experience in mid-stage technology development spanning both industrial and digital domains, including 10+ years developing digital twins in the aviation, energy, oil & gas, and renewables industries. Dr. Rasheed obtained his Ph.D. from Caltech where he studied experimental hypervelocity aerothermodynamics (orbital reentry heating). Recognized by MIT Technology Review Magazine as one of the “World’s Top 35 Innovators”, he was also awarded the AIAA Lawrence Sperry Award, an industry award for early career contributions in aeronautics. He has 32+ issued patents and 125+ technical publications relating to industrial analytics, operations optimization, artificial lift, pulse detonation, hypersonics, shock-wave induced mixing, space medicine, and innovation.