Cooperation Driver - Autonomous Intelligent Vehicle - CoCoVeIA

The ANR CoCoVeA project (2013-2017) has laid the foundations for a cooperative system for the automated vehicle (level 2).

Building on its results in shared driving, the CoCoVeIA project aims to integrate self-learning capabilities into the system, giving it the ability to analyze and understand the driver's actions during shared and manual driving phases, to achieve two key objectives:

1. The system, by learning about simple situations from the actions performed by the driver, will adapt its behavior to the driver's preferences and thus improve acceptability and confidence.
2. The system, by learning on complex situations (at the tactical level), will extend its skills and thus become, over time, capable of helping the driver in more varied situations.

The approach we propose consists in enabling the system to learn how best to execute well-defined maneuvers, modeled beforehand in a deterministic way (for compliance with traffic rules and safety) in a driving context shared with the driver (to improve efficiency and acceptability). To this end, the present submission proposes solutions for increasing the skills of driving automata via "progressive learning" properties of automata, learning made possible by the implementation of continuous sharing of vehicle control.

The objectives are manifold:

1. Foster the development of automatisms by "mimicking" human behaviors, while ensuring that these behaviors are compatible with safety;
2. Enable drivers to build their trust in the vehicle via their driving experience;
3. Personalize as much as possible the operation of automated systems according to the wishes of each driver;
4. Put automated driving systems on the road that integrate harmoniously with traffic.