DSpace Launches Simphera AV Development Simulator
The new cloud-based and scalable software is engineered to simulate and validate autonomous driving applications.
Already a well-known provider of hardware-in-the-loop (HIL) and software-in-the-loop (SIL) systems, dSpace has launched a new scalable, cloud-based simulation platform called Simphera, designed to speed development of autonomous vehicles (AVs) by managing computation-intensive validation of autonomous driving functions. Its web interface and cloud architecture are geared towards widely scattered global development teams and the sim features reusable models and scenarios to permit continuous testing on SIL and HIL platforms.
The Paderborn, Germany-based dSpace is responding to expanding need within the AV development community to help manage the number of tests that require large and scalable computing infrastructures. The new sim platform also serves the increase in virtual tests frontloaded into earlier phases of AV development.
“Simphera will make life easier in all phases of development and validation – from the initial idea of an algorithm to the release test of an ECU,” said Tino Schulze, executive VP of automated driving & software solutions at dSpace. “By providing this integrated solution, we not only accelerate development, but also make it more efficient, thereby increasing our customers’ return on investment.”
Prepare, simulate, validate
The Simphera simulation platform is based around three key modules. The first, Prepare, provides the tools needed to generate and modify artifacts required for development or validation tasks, including sophisticated (or dedicated) sensor-, road- and traffic-simulation models. The Simulate component brings the platform model together with the controller in specific scenarios within interactive simulations, and lets users check algorithm behavior via visualization and measurement functions. The third component, Validate, helps users perform scenario-based testing on a large scale.
The initial Simphera release is focused on providing validation solutions for autonomous-driving functions, scenario-based testing and SIL testing. According to dSpace, its reusable models permit a more seamless transition to automated HIL tests, and the traceability in troubleshooting reduces cost while speeding development homologation. “Everybody knows dSpace from hardware, because we are one of the pioneers in HIL,” explained Michael Peperhowe, lead product manager at dSpace for simulation models and scenarios, automated driving & software solutions. “But now it also gets more and more into the SIL environment.”
“Usually, our customers already have some infrastructure they would like to reuse, or at least partly reuse,” Peperhowe continued. “Therefore, it is important that Simphera is not just a ‘black-box,’ but an open framework. It brings everything out of the box. This is what many customers also like.”
Meeting AV development demand
According to Peperhowe, a new platform like Simphera was needed to help development engineers meet the demands of increasing test complexity. “Especially at these SAE [automated-driving] levels, we have different levels of complexity and driving functions, from active safety until autonomous driving,” Peperhowe noted. “AV development is one of the most complex tasks. Multiple smart systems have to work together with the utmost reliability. For the development and validation of such systems, the newest technology is needed.”
“To bring such systems into homologation, on the roads, it will not work with [physical] prototypes only. You need simulation. You need to drive millions of miles per night,” he continued. “This was the motivation to develop something new that meets these requirements for different driving functions. Create thousands of tests with just a few clicks and scalability in the cloud. You do not have to wait step-by-step until the results are ready. You can have a large number of simulation runs in parallel, which saves real time.”
Moving to the cloud lets customers choose how much of the computing assets they wish to dedicate to the simulations – often crucial in terms of meeting homologation milestones. “For sure, we need a flexible platform for doing [these] really huge amounts of tests. That means you need something to run in the cloud,” Peperhowe said, pointing out that by its nature, the cloud also enables global collaboration.
Automating edge cases
Simphera’s simplified interface is designed to let engineers quickly generate variations on standard testing regimens such as NCAP (New Car Assessment Program), or craft edge-case scenarios to validate less-common parameters. “This is quite important. Let's say the customers want to do homologation to get an approval of their controller for cars on the street. This will not work without qualified simulation models and tools behind that,” Peperhowe explained. “To generate scenarios from scratch, we have interactive editors available,” he added.
“Physics-based sensors are needed – camera, radar, lidar and so on – so you have different levels of test depth that you could take into account,” Peperhowe qualified. “The full AV stack, just the sensor perception or the pure ADAS algorithm. Even if everybody's talking about ADAS, often it's forgotten to take the powertrain into account. If we are combining e-mobility, [energy] recuperation, fuel-cell technology, etc., it has big benefits if you have models for all of these. That means you can use it for single-component tests or for full integration tests.”
Its experience in hardware testing likely provides dSpace an advantage in simulation software development. “We know what needs to be tested because of our long history, especially from the HIL side. Now, in addition, everybody wants to bring it into a virtual environment, front-loading to an earlier development stage,” Peperhowe claimed, emphasizing the flexibility of the cloud platform. “It's completely scalable by the user. You can start it on Friday before going home and come back Monday. Or, if you are in a hurry – ‘I want to have results within half an hour,’ – just scale it up. The limit is not on the dSpace tooling side, it's on the infrastructure from the cloud side. [Engineers] do not have to spend their brain about how to handle the tooling, they can concentrate on their work.”