Combining gestural and contact information for visual guidance of multi-finger grasps