Essais & Simulations n° 138

DOSSIER AUTO MOTIVE

DOSSIER AUTO MOTIVE INTERVIEW Towards safety and performance criteria similar to those of aeronautics ln this special report on the automotive sector, Essais & Simulations has chosen to focus on the autonomous vehicle and to interview several major publishers of digital simulation software, starting with Serge Laverdure, Connected & Automated Vehicle Unit Director at ESI Group. HOW DOES THE AUTONOMOUS VEHICLE REVOLUTIONIZE THE AUTOMOTIVE INDUSTRY? Driven by the shifting of end-users' expectations, the automotive sector faces one of its biggest challenge: the evolution towards a new concept of mobility; from product consideration to usage one. It's the outcome revolution. The autonomous vehicle is a consequence in term of product of this revolution. End-users are looking for new mobility solutions: environmental-friendly, autonomous, shared, etc. The impact of this evolution is multiple for manufacturers. Firstly, to change they need to innovate while continuing to produce their current products and solutions, key for their current growth and profitability. They need to invest in their innovation while reducing their overall costs and their time to market. Secondly to create the change, they need to modify, to redefine their engineering and manufacturing processes and methodologies. WHAT ARE THE IMPACTS ON MANUFACTURERS' ACTIVITIES IN TERMS OF DESIGN AND TESTING? System complexity, safety and acceptability are the most impacting aspects coming with higher level of automation. Classic Serge Laverdure Connected & Automated Vehicle Unit Director validation based on a mixt between system tests in the laboratory and real test on the road is no more possible because too costly and incompatible with pressure of time. Therefore, virtual testing is mandatory in order to address by simulation either critical scenario which are not reproducible with physical test or scenario performed under diff"erent weather conditions (rain, fog, snow ... ) and taking care with some key material properties of the surrounding. To summarize, to assess the reliability of an autonomous vehicle, there should be real data representing the many scenarios it is likely to encounter. These data are missing, and it is impossible to cover the millions of kilometers that would allow them to be collected. The design of an autonomous vehicle will therefore necessarily involve virtual prototyping and simulation. HOW CAN DIGITAL SIMULATION MEET THE CHALLENGES OF AUTOMAKERS ? Simulation should meet such new challenges but some pre requisite must be answered. The time being, on the one hand, automotive industry is using detailed simulation during the design of the product but with off line solutions (it means solutions without assessment in real time). Such solutions are related to one domain and not scalable to address complex systems. On the other hand, automotive industry is using driving simulator during the validation of the final product, more or Jess real time, but with very basic physic. These driving simula tors are far from the needed level of accuracy for covering ail topics about performance or safety. Therefore, simulation could meet the challenges of automakers if virtual test approximates the real test that should be very close to the operational do main of the vehicle. Having simulation dealing with physics-based models at ail stage of the systems is mandatory 56 IESSAIS & SIMULATIONS• N ° 138 • Septembre - Octobre 2019

DOSSIER AUTO MOTIVE and such models should be also qualified. About the challenge to innovate while continuing to produce, the use of physics-based simulation allows automakers to unleash their innovation stages while keeping the production rate and performance level. AS A SIMULATION SOFTWARE PUBLISHER, HAVE YOU ALREADY DEVELOPED SOLUTIONS THAT ADDRESS THE ISSUES RAISED BY THE AUTONOMOUS VEHICLE ? As a driver, you need to see, to analyze the situation you are facing and according your experience, concentration and the emergency of the situation, care with the driving task that include breaking, cutting, overtaking .... But as you have already experienced in your driver life the situations are always different and depending of certain conditions very difficult to anticipate. We have developed solutions in order to virtually replace the perception part which is achieved with camera, radar, Lidar. .. in the case of autonomous vehicle. We have the best of the class sensor mode! because we are taking care with all physic relevant aspect that impact the detectability of objects. Therefore, our customers can address performance and safety issue with our solution. WHAT ARE THESE ISSUES, BOTH IN TERMS OF THE SAFETY OF THE DRIVER, PASSENGERS AND OTHER ROAD USERS ? with its environment: other vehicles or infrastructure. This is the V2X communication that complicates this stage of safety even more. To obtain such a safety certification, the industrialization milestone can only be reached through the use of simulation and virtual prototypes. IN ADDI TION TO SAFE TY, WHAT ARE THE OTHER CHALLENGES OF THE AUTONOMOUS VEHICLE THAT DIGI TAL SIMULATION CAN HELP TO MEET? One of the other issues linked to Autonomous Vehicle is the end-user acceptance. It's an aspect related to the cognitive aspect of the human being. Indeed, social problems related to public acceptance, user awareness and ethics have become priority concerns for manufacturers and public administrations. Sorne projects aim to improve the response and sensitivity of the autonomous vehicle, making it more aware of the occupants, pedestrians and other drivers needs. ESI is part of a European Project named SUaa VE (SUpporting acceptance of automated vehicle): https://trimis.ec.europa.eu/project/supporting-acceptance-automated-vehicle. • Comments collected by Olivier Guillon Today, any regulation exists to certify autonomous vehicle and to set-up a framework with safety requirements both in term of safety for the driver, passengers and other road users. The trend strives for safety and performance criteria similar to those of aeronautics. It should be possible to know and certify the vehicle's reaction in ail possible situations (related to mechanics, electronics, driver, passenger, other vehicles, infrastructure, weather, etc.). A physical simulation of ail these scenarios is impossible. The last problem related to safety is the integration of considerations related to the interaction of the autonomous vehicle ESSAIS & SIMULATIONS• N ° 138 • Septembre - Octobre 2019157