For the development and the diffusion of vehicles equipped with autonomous driving systems, the safety matter is obviously central. The results of the research “Global Automotive Consumer study” by Deloitte have highlighted, for instance, that European consumers feel safer on an autonomous-driving car produced by manufacturers of the automotive industry than in cars proposed by hi-tech companies, thus revealing people’s trend to perceive brands already known and “tested” as more reliable. The safety of autonomous driving systems is a topic to which TÜV SÜD (independent body that operates in certification, inspection, testing, tests and training) has dedicated the “Driver Assistance Systems Conference”, held in Munich, in Germany, with the participation of over 200 engineers. Among the themes treated – as underlined Pietro Vergani, Business Unit Manager Consumer Product of TÜV Italia – they debated also the simulation issue, which will play a fundamental role in the future test and homologation activities. «Using these methods – he stated – test structures can assess daily up to 50 million scenarios: an unfeasible enterprise with physical tests on vehicles». Another subject discussed in Munich concerned the electronics of cars and these systems’ capability of imitating human behaviour.
The engineers attending the conference mentioned “neural networks”, i.e. a sort of artificial nervous system that allows machines or computers to learn autonomously, in compliance with the artificial intelligence vision. Besides, at the conference they presented IMAGinE project, which develops advanced driving assistance systems able to introduce support actions to the autonomous driving, too. Concerning instead platforms, the representative of an automotive company attending the conference underlined that “non-exclusive” ones will become more common in the future, they will be at disposal of various automotive companies and this will result in a reduction of costs, meanwhile making available a growing number of vehicles. Finally, in the opinion of Pietro Vergani, innovations also in car homologation methods will have to correspond to technological innovations. «We have analysed thoroughly new possible homologation approaches for autonomous-driving cars – explained Vergani – and we have inferred that virtual homologation methods, based on test simulation of virtual vehicles, are already reality and transposed by the legislation in course, even if not extensively and for dedicated approval ambits, like for instance the simulation of mechanical stresses on frames or underrun bars. Such methods, and the relative technical validations, are steeply rising and they will be the control basis for the vehicles of the next future».