Grasping is a key capability in robotics: the gripper is at the interface between the robot and its environment. The problem of versatility, i.e. the ability to adapt to many grasping tasks, is an active subject of research. For grippers, multi-digital and under-actuated architectures have been designed with this objective. Concerning the versatility of the grasp planning, the algorithm must be able to adapt to several objects and to determine, among a set of possible grasps for a given object, the grasp to execute according to the context. To do this, this set of possible grasps must be determined and characterised beforehand in a grasp space exploration phase. In the case of a multi-digital and underactuated gripper, this exploration is m...