Embodied intelligence has witnessed remarkable progress in recent years,<br>driven by advances in computer vision, natural language processing, and the<br>rise of large-scale multimodal models. Among its core challenges, robot<br>manipulation stands out as a fundamental yet intricate problem, requiring the<br>seamless integration of perception, planning, and control to enable interaction<br>within diverse and unstructured environments. This survey presents a<br>comprehensive overview of robotic manipulation, encompassing foundational<br>background, task-organized benchmarks and datasets, and a unified taxonomy of<br>existing methods. We extend the classical division between high-level planning<br>and low-level control by broadening high-level planning to include language,<br>code, motion, affordance, and 3D representations, while introducing a new<br>taxonomy of low-level learning-based control grounded in training paradigms<br>such as input modeling, latent learning, and policy learning. Furthermore, we<br>provide the first dedicated taxonomy of key bottlenecks, focusing on data<br>collection, utilization, and generalization, and conclude with an extensive<br>review of real-world applications. Compared with prior surveys, our work offers<br>both a broader scope and deeper insight, serving as an accessible roadmap for<br>newcomers and a structured reference for experienced researchers. All related<br>resources, including research papers, open-source datasets, and projects, are<br>curated for the community at<br>https://github.com/BaiShuanghao/Awesome-Robotics-Manipulation.