“We’re now more efficient than before by a factor ranging between two and three”
Interview with Christoph Kossira and Dr. Paul Spannaus of Elektronische Fahrwerksysteme GmbH (EFS)
The networking of vehicle systems has become increasingly extensive in recent years and a change in this trend is not on the horizon. EFS, a joint venture with Audi AG, was seeking a solution to virtually validate ESC (electronic stability control) systems and to thus enable a significantly more efficient development process. We talked to Christoph Kossira and Dr. Paul Spannaus about the successful switch to a new toolchain that was accomplished based on the open integration and test platform CarMaker.
At Apply & Innovate 2016, you gave a presentation on the benchmark of simulation solutions. Could you provide a brief outline of this once more?
Kossira: We reported about the establishment of a new toolchain in our presentation because three years ago we were commissioned by Audi to virtually validate ESC (electronic stability control) systems. The toolchain that existed at the time was past its prime, so there was a need for us to make some changes.
Spannaus: The idea was to develop our desired test environment using an example project with an example test scope and to subsequently run tests in these diverse simulation worlds utilizing the tools.
From individual tests to virtual approval – what elements do you accomplish with virtual test driving?
Kossira: We take care of the process steps that by and large are no longer carried out in the vehicle. Clearly, our goal is to perform more and more maneuvers from previous real-world road testing with virtual test driving. This also includes making recommendations for approval by means of virtual testing. Here’s a case in point: In the area of stopping and deceleration management, there are tests which are only feasible in the virtual world because a vehicle’s response cannot be controlled with accuracy down to twenty milliseconds in real-world road tests.
Spannaus: In any event, we need virtual tests for validation because this is the only way in which we can truly test all scenarios reproducibly. Naturally, a lot of things are still done using real-world test drivers. They must have the requisite skills so that driving maneuvers can be carried out identically in each test.
How do you assess this for the area of ADAS? Is it absolutely necessary to have validated virtual prototypes in this field?
Kossira: It depends on what you want to do. For a classic ACC, it’s probably not absolutely necessary. However, as soon as it comes to functions such as emergency braking and collision avoidance, vehicle dynamics must be properly modeled.
This is also about providing developers with turnkey models, in other words virtual prototypes, with exactly the required level of detailing. It must be possible to adapt the model to the task.
Especially in the field of automated driving there is still a relatively large number of open issues. In your view, what are the key issues to focus on?
Kossira: The area of automated driving is extremely extensive. I believe that those who will be able to validate this entire area will be the winners. However, at the moment, no-one is able to fully validate it. Everyone knows about the infinitely large variety of events. But who defines the events that are important? No matter what, this will require a departure from manual labeling and increasingly happen by means of automation. The ability to validate such driving functions will still pose a truly great challenge to all of us.
How will this topic continue to develop from your perspective?
Kossira: It plays an increasingly important role day after day. Due to diverse inquiries, the focus becomes sharper because they result in specific requirements. Fully automated driving across the board will not become feasible at a moment’s notice. Just imagine some of the European roads that suddenly widen from three to five lanes. By contrast, road traffic in other countries is managed in completely different ways, or with respect to the pace in the Asian region for instance, I can imagine that lanes specifically dedicated to autonomous vehicles might be built.
This means that, in your view, simulation is gaining more and more recognition?
Kossira: Yes, it’s gaining more and more recognition. Plus, simulation is clearly better than it used to be. You no longer have to be an expert to be able to operate something like this. Even laypersons can learn it relatively fast. The results you obtain are relatively comprehensible, good and comparable.
In retrospect: Would you say that your project was a total success?
Spannaus: Yes, on closer reflection: We’re now more efficient than before by a factor ranging between two and three – and with the same number of employees involved.
Kossira: I agree. It’s a success for our internal mindset as well. Our claim is to look ahead. We seek to implement ever new technology. In retrospect, we can definitely say that now we’re capable of a lot more than we were last year.