目的 介绍机器人无序分拣技术最新进展和研究成果，为复杂场景下的机器人自动化应用提供研究思路和技术支撑。方法 从机器人无序分拣过程中的3个关键技术任务展开论述，具体包括散乱目标的检测与识别、目标的空间姿态估计与无序分拣的抓取决策，并对各个任务中涉及的方法进行分析总结。结论 机器人无序分拣技术目前面临的主要挑战在于处理复杂环境下散乱堆叠物体的视觉感知、位姿估计和抓取决策。具体的分拣系统应当考虑实际的场景环境，结合任务需求进行设计，以求达到机器人尽可能替代人力劳动的目的。

This paper introduces the latest progress and research results of robotic unordered picking technology, and provides research ideas and technical support for robot automation application in complex scenarios. It discusses three key technical tasks in the robot unordered picking process, which including detection and recognition of cluttered objects, spatial pose estimation of objects, grasping strategy of unordered picking. Analyzes and summarizes the methods involved in each task. The main challenges of robotic unordered picking technology today are dealing with visual perception, pose estimation and grasping strategy of cluttered and stacked objects in complex environments. Specific picking systems should be designed considering the actual scenario environment and combined with the task requirements in order to achieve the goal of replacing human labor by robots as much as possible.