Cooperation and learning to increase the autonomy of ADAS

This paper discusses on the cooperation and the learning processes to increase the autonomy of a human-machine system or an artificial or human agent. The autonomy is defined as the capacity for a system or an agent to fend alone. It is described in terms of competences and the limits of these competences. Cooperation and learning aim then at increasing the competences or managing the system limits. The management of the autonomy is detailed through different structures of cooperation. It concerns the sharing control between systems or between agents in order to recover their limits. Different classes of learning processes are proposed: the mimicry-based approaches, the dysfunction-based ones, and the wait-and-see-based ones. Advanced Driver Assistance Systems (ADAS) are usually designed integrating cooperation characteristics. Two case studies about the use of cooperative ADAS are then proposed. They are hypothetical scenarios that are discussed to introduce possible future ADAS perspective implementing competences such as learning or cooperative learning.