The Dial-a-Ride Problem (DARP) models an operation research problem related to the on demand transport. This paper introduces one of the fundamental features of this type of transport: the robustness. This paper solves the Dial-a-Ride Problem by integrating a measure of insertion capacity called Insertability. The technique used is a greedy insertion algorithm based on time constraint propagation (time windows, maximum ride time and maximum route time). In the present work, we integrate a new way to measure the impact of each insertion on the other not inserted demands. We propose its calculation, study its behavior, discuss the transition to dynamic context and present a way to make the system more robust.