Precision grasping in a visual form agnostic