Robotic manipulation using robotic arm enables robots to assist humans in completing a variety of tasks that are complicated and risky, such as picking boiling hot bowls and placing them on the dining table. Robotic manipulation includes three processes, taking the photos by the camera installed on the robotic arm, calculating the optimal grasp poses and controlling the robotic arm to grasp the objects, and the second process calculating the best grasp poses is the most complicated. Since the second process is the most complicated, this paper aims to conclude the whole process and related methods of the grasp poses generation. This review paper divides the grasp poses generation into three subtasks, which are object identification, object poses generation and grasp poses generation. Meanwhile, this review paper provides deep insights into each subtask of the grasp poses generation. The purpose of this review is to provide academic and research direction for future research in robotic arm autonomous grasping.