Conclusion, challenges, and future avenues for development

Perception of the road and urban environment is an essential component of driving, whether human or automated. While advanced driver assistance systems (ADAS) have proven effective in assisting drivers with this complex task, experiments on test tracks and in real-world conditions show that the level of operational automation remains low, particularly in terms of safety and traffic optimization. The many challenges to be overcome mainly concern the perception of the road environment, an area where progress remains insufficient given the diversity of situations encountered.

Indeed, the road environment (interactive, complex, and dynamic) is composed of five classes of actors (obstacles, road, ego-vehicle, environment, and driver), whose attributes must be accurately estimated in order to build reliable, robust, and accurate local dynamic perception maps that feed...