Connected and Autonomous Vehicle. Technologies, Challenges and Deployment

The concept of the autonomous vehicle dates back to the 1980s, when it first aroused enthusiasm in research and industry circles, followed in the early 2000s by a period of wilderness, mainly due to major uncertainties about its acceptability to users. The arrival of the Google Car changed all that, creating a real media shock rather than a technological one, in a context that had become favorable due to the growing and generalized pressure of automated enforcement. Since then, interest in autonomous and connected vehicles has continued to grow, and the emergence of a market has thrown all manufacturers and equipment suppliers into the race. This led to a clear, consensual definition of the concept in the form of five levels of automation, ranging from simple driver assistance systems to full automation where the presence of a driver is no longer necessarily required. Thanks to this graduated approach, the next two decades will see more of an evolution than a revolution: advanced driver assistance systems (ADAS), at levels 2 and 3, contain the technological building blocks that will constitute the key functions of autonomous vehicles at levels 4 and 5, the most important of which are : precise localization, obstacle detection, vehicle dynamics modeling, analysis and understanding of the road scene, trajectory planning, situational awareness and driver status monitoring. All these building blocks must be integrated into the vehicle to form the on-board electronic and computing architecture of autonomous vehicles. For a long time, autonomous vehicles were thought to be a purely "automotive" affair, but the end of the 2010s highlighted the importance of infrastructure and underlined the synergies to be brought into play in a win-win collaboration between the automotive and road industries and public authorities, all with a view to increasing safety and optimizing mobility. However, not all the technical difficulties have been resolved: in fact, they have been considerably underestimated, making it very difficult to forecast the future, especially for level 5 autonomous vehicles. In order to identify and reduce this complexity, it has become clear that it is necessary to better define the situations in which autonomous vehicles will be required to evolve: these are the recent notions of ODD (Operational Design Domain), which sets the framework for vehicle evolution, and ISAD (Infrastructure Support levels for Automated Driving), which specifies the expected contribution of the infrastructure. In Europe, all these advances have made it possible to clarify the progressive deployment of autonomous vehicles according to technological targets (types of vehicle) and use cases (scenarios of use), which have been specified by the European ERTRAC platform. But beyond the use cases, new mobility practices are taking shape worldwide,...