Simulation environments that replicate situations for autonomous driving are like a virtual planet on which a vehicle has to master innumerable traffic situations. It is just such a digital world that the SIMPHERA software provides, adding to the already broad range of dSPACE products for simulation and validation.
It is Tuesday morning. Function developer Belle Fisher powers up her computer and accesses her company’s simulation environment via the web browser. The environment notifies her about a new test scenario that is planned and available, which her colleague Yang Yu in China has prepared. The scenario is ideal for evaluating the algorithm for highly automated driving that Belle and her team are currently working on. So she integrates it into her simulation, uses it to test the algorithm, and optimizes the algorithm based on the insights gained from automated validation. The result is a validated new version of the algorithm.
Later in the day, sensor specialist Roni Cohen in Israel integrates the model of a new radar sensor into the simulation environment. The sensor was selected as an alternative on the basis of the manufacturer’s data, to improve steering behavior in specific corner cases. Belle can immediately equip her virtual test vehicle with the sensor model and select the relevant corner cases as test cases, which she then uses for simulations testing whether the desired improvement is really achieved.
At the team meeting the next morning, the vehicle control project manager asks for an integration test to be run to investigate the behavior of the whole vehicle including the new sensors. So Belle runs a simulation with over 1,000 test cases, all generated by automatic parameter variation. And thanks to multi-instance simulation in the cloud, the development team gets the results just after lunchtime. They can now be compared with the results of previous simulations to identify any changes.
With the new improvements, the vehicle control meets its defined requirements. The team now has a newly validated version of the vehicle control.
Following successful validation by software-in-the-loop (SIL) test methods, the team tackles the next validation step: determining how the vehicle control behaves under real-time conditions, using the actual hardware. A service provider is tasked with implementing the software on the ECU, which is being developed in parallel. To test the ECU under real-time conditions, the service provider uses the same simulation environment as the development team and accesses the same test suites to select suitable test scenarios. In SIMPHERA, the service provider selects a hardware-in-the-loop (HIL) simulator as a test platform. This degree of integration, with test artifacts that can be directly reused, enables an efficient and reliable validation process.
If you can relate to any or all of these cases, our new cloud-based solution for simulation and validation is exactly what you need. Read more to discover what SIMPHERA can offer for your own tasks.
SIMPHERA is a web-based solution that offers flexible user access and exciting technologies.
In short: It is a completely new concept, a web-based cloud software solution for simulating and validating autonomous driving.
SIMPHERA has been specifically designed with the aim of getting your autonomous driving innovations on the road faster. The software brings them to life at an early stage and helps you analyze them and handle their complexity. How? By giving you easy access to powerful functions for simulating the vehicle dynamics, traffic, and more. So you can trial new algorithms, such as for vehicle control, in a virtual world: simpliCity. Developers can do this by completing reproducible, deterministic test drives with a virtual vehicle in a virtual environment. The vehicle, the environment, and the tests—or traffic scenarios as they are called in this context—can all be freely defined and imported. The virtual world of simpliCity is so versatile that any desired corner cases can be replicated to guarantee broad test coverage. And because the cases are easy to parameterize, new variants can be generated to increase test depth. After each simulation run, users are given a detailed and informative analysis of the vehicle behavior. For example, you can identify the speeds at which emergency braking leads to a collision.
Validating functions for autonomous driving with integrated scenario-based tests.
SIMPHERA gives you an easy and intuitive way of creating, performing, and evaluating highly realistic simulations. Its user interface provides role-based user guidance:
This structured procedure guides you through a workflow that avoids errors and enhances efficiency. Moreover, the highly integrated solution ensures that everyone works on the same data, so nothing “falls through the gaps” between one team or person and another. It is also transparent, letting everyone involved see the various work products.
SIMPHERA’s integrated workflow supports efficient end-to-end validation.
Wherever the users and the IT infrastructure might be located, SIMPHERA is there, too—as a web-based solution in the cloud or on your company’s server cluster, it can always be accessed directly. Meeting the needs of every user and ideal for worldwide, shared use. There’s no lengthy installation work, you only need the license and off you go. You can set up distributed teams and implement efficient processes, then collaboration can commence.
With simple, web-based access, SIMPHERA is ideal for distributed development.
Scenario-based testing means that particular driving situations can be generated automatically, with specific traffic constellations and unusual road user behavior that push the driving functions to their limits. The situations are replicated in the virtual test environment and can be run through repeatedly to test the driving functions in complete safety. These scenarios can also be abstracted to make them parameterizable, so innumerable “concrete” scenarios can be generated from one “logical” scenario via parameter variation – to further increase test coverage. SIMPHERA has a library of relevant driving situations (Euro NCAP, ALKS, etc.), which can be extended in whichever way required.
The basis for simulation is test scenarios that are either synthetically defined or generated from real sensor data and imported.
To be absolutely certain that driving functions perform safely in any situation that might occur on the road, the volume of testing required can easily get out of hand. Especially when the number of test cases and variants skyrockets, SIMPHERA shows its strengths. For example, tests can be scaled by means of parallel execution in the cloud. And even if extra work is required for test coverage and test depth, you can still keep your projects on schedule. With the support of SIMPHERA, you can uphold the quality and completeness of validation at all times, thereby ensuring that the algorithms are reliable and the vehicles safe.
Scalable tests in the cloud for faster results and broad coverage.
Once the setup is ready, the process is simple, and any function developer can use SIMPHERA to test their algorithms quickly. The Linux Docker containers are automatically orchestrated in the background by using Kubernetes (a system for automating the deployment, scaling, and management of containerized applications). After that, the individual simulation jobs are automatically assigned to the execution nodes. The user does not have to intervene. Once the jobs have run, SIMPHERA provides a clear overview of the results.
Once the user has selected a cloud platform, SIMPHERA takes care of the entire cloud-based test execution.
The validation of ECU software begins with early software-in-the-loop (SIL) tests and continues on through to hardware-in-the-loop (HIL) tests that check and validate the behavior of the ECUs and the vehicles under real-time conditions. SIMPHERA gives you access to both of these test platforms. This provides a fundamental benefit: Test artifacts can be reused for both SIL and HIL—with no need for conversion between different platforms. This reduces errors, saving both time and costs. It is also a fundamental precondition for traceability and homologation.
And last but not least, having one single integrated environment for all validation activities enhances development efficiency. Thus, SIMPHERA is an end-to-end solution for simulating and validating ECU software and ECUs.
End-to-end solution for simulation and validation: SIL and HIL tests with reusable models, scenarios, tests, and interfaces.
SIMPHERA’s philosophy is to ensure coverage-driven validation with measurable reliability. It does so by providing vital features such as mass validation and robustness tests. The industry-proven technologies in SIMPHERA, such as validated simulation models and established simulation platforms, ensure high reliability, particularly when it comes to validating that the software under test functions correctly.
Consistent data management, even between SIL and HIL tests, is an elementary prerequisite for traceability and homologation. In addition, SIMPHERA is open and flexible, letting you integrate partner solutions from the field of homologation.
Test reports show the simulation results.
SIMPHERA is being developed in close cooperation with pilot customers. This ensures that it has precisely the right features that the industry urgently needs when validating driver assistance systems and functions for autonomous driving. And there’s already a full agenda for further functions. The next items on the list include deep integration of realistic sensor simulation, and additional test methods. dSPACE would like to invite you, as a customer, to join in the agile development of SIMPHERA and help ensure that your requirements and wishes are completely fulfilled.
Opinion of a German OEM
Opinion of a Japanese OEM
Publié : 14 septembre 2021
S'inscrire à la Newsletter
S’abonner à nos newsletters, gérer ses abonnements ou se désabonner