AWS RoboMaker Simulation
Build Intelligent Robots Faster with RoboMaker Simulation
Autonomous robots require sophisticated software for capabilities such as navigation, perception, and object manipulation. This software requires comprehensive testing to ensure robots operate correctly in their environments.
However, increasing the number and variance of testing scenarios can be challenging. Testing software only with physical robots is manual process that slows down development and requires expensive hardware for each developer to test code with. Testing with computer-based simulation can also be difficult scale and manage, and can require large budgets and specialized skills.
AWS RoboMaker Simulation makes it easy to test robot applications, so you can bring high-quality robots to market faster. With RoboMaker, developers can run hundreds of varied tests using virtual robots in a 3D simulation worlds. RoboMaker makes it easy to create simulation worlds and run parallel simulations in large batches. By running tests in RoboMaker Simulation, teams can quickly find and fix bugs before they reach the robot hardware. This reduces code errors and speeds up development velocity.
Increase speed and code quality
AWS RoboMaker Simulation enables you to easily run hundreds of varied test scenarios, making it possible to increase testing coverage and quickly resolve errors before problematic code reaches the robot. By running tests in simulation, you can identify edge cases and focus real-world testing only where it’s needed most. Comprehensive testing in simulation helps improve the quality of your robot and ultimately, speed up development velocity.
AWS RoboMaker Simulation enables you to integrate your robotics testing into an automated testing and release process, allowing you to rapidly deploy new features. With RoboMaker, you can automate batch simulations after each code check-in and for nightly integration testing. Automating your test and release process allows you to quickly and easily test each code change and catch bugs while they are small and simple to fix.
Fast, easy simulation world creation
AWS RoboMaker makes building simulation environments faster and affordable by providing simulation worlds and a variety of 3D assets. You can use these simulation assets to build simulation environments to match the level of fidelity you require. Additionally, AWS RoboMaker WorldForge enables developers to generate hundreds of user-defined, varied 3D indoor residential simulation worlds in minutes.
Reduce hardware dependency
AWS RoboMaker Simulation reduces the need to test software on expensive robot hardware and eliminates the need to provision, configure, or manage any infrastructure. RoboMaker Simulation provides fully managed compute infrastructure for executing your robot code in natively supported ROS environments and simulation code in natively supported Gazebo environments.
Automated regression testing
Bugs and compatibility issues can be introduced with each code commit. It can take lengthy testing cycles to identify bugs, and they can be missed until late in development or in production when they are costly to fix. With RoboMaker, robotics companies can automate regression testing within a continuous integration and continuous delivery (CI/CD) pipeline to test quickly and at scale. Development teams can run batch simulations after each code check-in, for nightly integration testing, and before each software update is released.
Multi-robot fleet testing
Robots in a fleet need to be tested together in realistic working environments to ensure they behave safely and correctly when they interact. Testing robot interactions on physical robots is a logistically difficult and time-consuming process. RoboMaker Simulation enables you to quickly and robustly test inter-robot communications and routing algorithms for multiple robots as if they are in the same 3D environment. With RoboMaker, teams can connect multiple concurrent simulations to their central fleet-management software for testing.
For some complex tasks in unpredictable environments, such as autonomous navigation and object manipulation, developers train machine learning models using reinforcement learning (RL). It’s nearly impossible to train RL models adequately in real-world environments because of the large number of iterative training cycles needed for a model to learn quickly. With RoboMaker’s scalable, simulations, developers can train their RL models with high volumes of iterative trials, enabling them to bring intelligent robots to market faster.
Organizations that use robots often customize their robot functionality by developing Android, iOS, or web applications that work with the robot’s software. These add-on applications must be tested with the robot’s software, but it is costly and often impossible to provide each developer a physical robot and space to test it in. AWS RoboMaker Simulation makes it easier for developers to test applications that work with new robots in sandbox environments so they can bring integrated solutions to market faster.
Learn how iRobot uses AWS RoboMaker Simulation to validate robot software builds in hours instead of weeks by running 40 automated tests on each code commit and more than 500 automated tests for each code release candidate.
Learn how Milvus Robotics uses AWS RoboMaker Simulation and the AWS provided open source Hospital World to eliminate obstacles to robotics testing and deliver products to market faster.
Learn how a Portuguese innovation institute, INESC TEC used AWS RoboMaker Simulation to set up a simulation-based regression testing CI/CD pipeline and reduced their testing time from 4 hours per test in the physical world to a couple of minutes with simulation.