Fully managed compute service enables complex, dynamic 3D simulations that model systems with more than a million objects, quickly predict real-world outcomes, and create immersive training spaces for customers

Duality Robotics, Lockheed Martin, and uCrowds are among customers and partners using AWS SimSpace Weaver

LAS VEGAS — (BUSINESS WIRE) — November 28, 2022 — At AWS re:Invent, Amazon Web Services, Inc. (AWS), an Amazon.com, Inc. company (NASDAQ: AMZN), today announced AWS SimSpace Weaver, a fully managed compute service that helps customers build, operate, and run large-scale spatial simulations. With AWS SimSpace Weaver, customers can deploy spatial simulations to model dynamic systems with many data points (e.g., traffic patterns across an entire city, crowd flows in a venue, or factory-floor layouts), and then use the simulations to visualize physical spaces, perform immersive training, and garner insights on different scenarios to make informed decisions. AWS SimSpace Weaver allows customers to run a simulation that has more than a million entities interacting in real time and create more sophisticated environments than previously possible, while reducing the time to simulation deployment from years to months. AWS SimSpace Weaver handles provisioning Amazon Elastic Compute Cloud (Amazon EC2) instances as well as managing the network and compute infrastructure, so customers can focus on designing and observing simulations in real time, across multiple live viewers. There are no upfront commitments or fees to use AWS SimSpace Weaver, and customers only pay for the amount of compute used. To get started with AWS SimSpace Weaver, visit aws.amazon.com/simspaceweaver.

Today, customers across industries such as transportation, robotics, and public safety need to create dynamic, 3D experiments to understand possible real-world outcomes and visualize and train in immersive spaces. For example, to realize the impact of road closures, a customer can simulate different closure combinations on their routes. These types of complex spatial simulations demand significant compute resources, stretching the existing limits of their compute capacity. Additionally, integrating and scaling a spatial simulation with millions of interacting objects across compute instances is a difficult process. It requires dividing the simulation—and its components—into pieces, keeping those pieces simulating at a consistent rate, and replicating relevant data across the environment as well as among multiple live users. Solving these problems can require years of investment and specialized hardware, resulting in simulations that are too costly for all but the most well-funded organizations. Further, spatial simulations must be programmed to capitalize on the size and density the additional compute enables, requiring specialized teams trained in compute configuration and utilization. As a consequence, customers make tradeoffs in the sophistication of their simulation logic, limiting them to only exploring basic scenarios, while more nuanced problems—particularly those that are sensitive and expensive—are merely estimated.

AWS SimSpace Weaver is a fully managed compute service that enables customers to run complex, large-scale spatial simulations where they can conduct experiments that closely approximate real-world systems and visualize immersive training spaces. AWS SimSpace Weaver supports more than a million moving objects like people, cars, traffic lights, roads and more, each with their associated behaviors and properties all interacting in real time. Like an actual city, the simulation is an expansive “world” in itself. To build simulation applications, customers can use the AWS SimSpace Weaver software development kit to incorporate their custom simulation content or use built-in integrations for third-party solutions like Unity and Unreal Engine 5. Once a customer is ready to deploy their simulation, AWS SimSpace Weaver automatically sets up the environment, connects up to 10 highly available, low latency Amazon EC2 instances into a networked cluster, and distributes the simulation across instances. AWS SimSpace Weaver then manages the network and memory configurations, replicating and synchronizing the data across the instances to create a single, unified simulation where multiple concurrent users can interact with and manipulate the simulation in real time. Customers can focus on designing simulations, observing the performance of simulations, and applying their learnings to prepare for scenarios and make business decisions. The scale and elasticity of AWS SimSpace Weaver—through its pay-as-you-go model—means customers do not need expensive hardware or additional headcount to manage spatial simulations. AWS SimSpace Weaver allows a variety of customers to easily run complex, immersive 3D experiences and test and visualize large-scale scenarios.

“Previously, if a customer wanted to scale up their spatial simulation, they had to balance the accuracy of the simulation with the capacity of their hardware, which limited the usefulness of what they could learn,” said Bill Vass, vice president of Technology at AWS. “AWS SimSpace Weaver removes the burden of managing simulation infrastructure, simplifying how customers run large-scale simulations and freeing them to focus on creating differentiated content and expanding access to simulation development. Whether the goal is to simulate the population of urban megacities, track logistics operations on a continental scale, or build real-world scale digital twins for smart city initiatives, AWS SimSpace Weaver empowers customers to create vast, distributed simulations.”

AWS SimSpace Weaver provides a local development environment for customers to iterate and test a version of their spatial simulation for free on their personal hardware, before running their simulation at scale on AWS. The local environment also uses the same APIs as AWS SimSpace Weaver, so customers can transition their simulation to the cloud without modifying any code. AWS SimSpace Weaver is generally available today in US East (Ohio), US East (N. Virginia), US West (Oregon), Asia Pacific (Singapore), Asia Pacific (Sydney), Europe (Frankfurt), Europe (Ireland), and Europe (Stockholm), with availability in additional AWS Regions coming soon.

Duality Robotics is building a digital twin integration platform for solving real world problems. “Until now, simulation has always involved a tradeoff between scale and fidelity,” said Apurva Shah, CEO at Duality Robotics. “The combination of Duality Robotics’ digital twin simulator, Falcon, with AWS SimSpace Weaver takes digital twins of environments and systems developed for the Unreal ecosystem and makes them instantly available for massive scale simulation. The integration with Epic Games’ Unreal Engine and access to a vast collection of Unreal assets powers realistic 3D environments and visualizes data with a high level of accuracy necessary for efficient simulation deployment. Duality Robotics’ emergency medical services use case demonstrates the value of spatial simulation with high-fidelity physics, photoreal rendering, virtual sensors, and real-time performance with over a million concurrent agents. This is proving customers can run AWS SimSpace Weaver to model sensitive physical systems and garner insights applicable to the real world.”

Epic Games is a leading interactive entertainment company and provider of 3D engine technology. Unreal Engine is Epic Games’ complete suite of creation tools for game development, architectural and automotive visualization, linear film and television content creation, broadcast and live event production, training and simulation, and other real-time applications. “While developers previously had to compromise on scale versus fidelity when creating simulations, they now can create environments that more accurately reflect the amount of people and objects in the real world,” said Nathan Thomas, vice president of Unreal Engine at Epic Games. “By combining the power of Unreal Engine 5 with AWS SimSpace Weaver, customers are able to scale our City Sample project to over one million interactive pedestrians. This opens the door for developers to create massive digital twins and simulations with new levels of realism.”

Lockheed Martin is a global security and aerospace company principally engaged in the research, design, development, manufacture, integration and sustainment of advanced technology systems, products and services. The Virtual Prototyping group of the Skunk Works division is chartered to provide advanced visualization, modeling and simulation, and human computer interfaces capabilities. "We need to be able to simulate at real-world scale to trust that the insights we gain from simulation are transferable back to reality. Historically, it would take a long time to simulate large scenarios and we were limited to only a few users that could interact with the simulation at the same time,” said Wesley Tanis, virtual prototyping engineer at Lockheed Martin. "We worked with AWS to develop a San Francisco Earthquake Recovery demo, which illustrates the ways first responders might organize an aid relief mission. Together, we were able to simulate over a million objects in near real-time at a continental scale, giving us real-world insight to increase our situational preparedness and planning across a wide range of scenarios including natural disasters.”

Maxar is a leading provider of comprehensive space solutions and secure, precise geospatial intelligence for government and commercial customers, helping them to monitor, understand and navigate our changing planet, provide global broadband communications, and explore and advance the use of space. “Typically, spatial simulations require designers and developers to build a world first, then augment it with assets to create a compelling and immersive world. Only once the world is complete, does the team consider the computing power required to run the simulation,” said Dan Nord, senior vice president and general manager of Enterprise Earth Intelligence at Maxar. “Now, with the 3D geospatial digital twin of Earth built with Maxar’s Vivid imagery basemap, blackshark.ai’s synthetic buildings and textures, and AWS SimSpace Weaver handling compute infrastructure management, designers and developers can jump straight into creating vast, detailed simulations that are ready to scale. In tests using AWS SimSpace Weaver with our 3D digital twin, customers have been able to build a real-world simulation faster than previous methods.”