Robotic coverage problem is addressed in several fields: painting, stripping, grist-blasting, etc. In multi-robot systems, the collaboration between robots reduces the cycle time and increases the coverage task accuracy. However, the robot base placement must be deeply studied to attend those goals. In this article, we propose a new approach in order to assign tasks within a multiple robots system. In addition, we develop an optimization strategy to find the optimal number of robots with their optimal poses required to cover the entire surface. To assess our method, our algorithm was tested on regular surfaces such as cylinder and hemisphere, and on a complex surface represented in a car shell.