Discover more about the topics and technologies to be discussed at this year's conference, via a series of exclusive interviews with a selection of our expert speakers
Ferenc Pintér, AiSim product manager at AImotive, outlines how the company approaches simulation for AVs and why it’s so crucial. He will talk more about this in his presentation at the Autonomous Vehicle Test & Development Symposium. For more information, up-to-date programs and rates, click here.
What are you currently working on?
I will discuss how automated testing with simulation tools can accelerate the development of automated driving solutions, while also making them safer. One of the founding principles of AImotive four years ago was to adapt aviation-grade safety in AV technology through the extended use of simulation. Use of simulation in commercial aviation is hugely widespread and the industry continues to report significantly fewer injuries or fatalities per mile travelled than road travel.
Thanks to the advanced simulation techniques that have been employed in aviation for decades, there are situations in which simulation is accepted as a complete proxy for real-world verification and validation. Although real-world testing will remain crucial for the safe development of cars that will eventually navigate our roads by themselves, it’s our fundamental belief that the auto industry must learn from aviation and strive to maximize the use of simulation.
From this belief sprung aiSim, our complex, end-to-end simulation tool. We started developing aiSim from the very early days of AImotive in an attempt to deliver a viable tool for auto makers to support the complete development process of automated vehicle technology, from software design to production. At this stage, we are engaging with global automotive brands who deploy our simulator for their respective technologies. We are also advancing aiSim with features that help our partners meet the safety requirements of ADAS and autonomous functions set by Euro NCAP.
How do you approach AV simulation?
One of the major advantages of simulation compared to real-world testing is the ability to focus on interesting miles, thus saving a considerable amount of time and money. In the virtual world you don’t need to wait for interesting test cases – or scenarios as we call them – to happen randomly. Instead, they can be created. A scenario is a predefined traffic situation where specific criteria which the system under test must meet to pass are set. The goal is to be able to measure the automated driving system’s ability to handle certain tasks. Scenarios can be permuted with different traffic densities, weather conditions, time of day, road surface and layout, to name but a few. aiSim comes with a standalone scenario editor to build new scenarios and templates.
We follow three unique approaches to create a varied database of scenarios for development and testing purposes. The first is that we create scenarios to verify a specific functionality with a well-defined set of requirements for that purpose. Take lane keeping assist. A simple scenario in this case would be selecting a challenging curvy section of highway and setting the requirement for the system to stay in the designated lane for 1km.
The second methodology we use is to create scenarios based on dangerous situations derived from the Euro NCAP and NHTSA road accident databases. These scenarios will prepare the software for situations that are common on our roads but don’t always get assessed by the usual functionality tests. Finally, we can draw scenarios directly from the real world. When a vehicle in our test fleet encounters a situation it cannot handle at first, this will be included in the simulation database. It is this ability – to take real-world disengagements and create virtual proxies to help solve them – that makes aiSim the center of our development pipeline.
What do you see as the ideal balance and relationship between simulation and real-world testing for AV and ADAS?
No one can avoid real-life testing due to its unsurmountable, vivid complexity. However, it can serve as a last-mile verification. For the vast majority of the journey, simulation continues to prove itself as an invaluable asset. With simulation, all existing sensor footage, incident reports and accident records are there for the industry to learn from. As such, trends show more and more collaborative approaches to share not just raw data, but also scenarios and interesting situations. As standardization and regulatory bodies catch up with the maturing industry that now places safety and public trust before reaching market first, the quality of AV and ADAS simulation tools is also going to reach new levels.
aiSim has been designed from the ground up to be a market leader in the era of these new tools. It facilitates automating solutions to help engineers quickly refine their ideas, and validate new release candidates over millions of virtual miles. By making tiny changes in each iteration issues can surface which would otherwise remain hidden. Simulation is also crucial once production is up and running, since each over-the-air update must also be equally reliable. Automakers therefore need solutions that allows them to swiftly react and improve their driving code every time new situations are encountered.
How is AIMotive moving simulation forward?
As we announced at CES 2019, aiSim comes with a purpose-built engine for testing automated vehicle technology, rather than a game engine. This is something that – despite the many limitations – much of the market still relies on. With the help of our unique development pipeline and large-scale testing automation infrastructure, we believe that our simulation approach has considerable economic benefits, helping our clients converge their R&D projects to solutions that are safe to productize. As ASAM (Association for Standardization of Automation and Measuring Systems) members, we participate in the development of multiple AV simulation standards, and we are continuously advancing our automated scenario creation and parameter permutation capabilities. As tool providers, our roadmaps are focused on the key challenges that the industry faces, such as data import, scaling, module integration and cooperative solutions.
Catch Ferenc’s presentation titled “Automating simulation for safer self-driving” at the Autonomous Vehicle Test & Development Symposium. For more information, up-to-date programs and rates, click here.
