智能机器人与人类智慧的融合，即人机协作共融，已经实现了将机器人的机械优势和人类的高级认知能力集中于同一个工作架构之中，能够在复杂环境中协同作业，从而提高效率。针对复杂的人机共融场景，特别是机器人在诸如光线条件变化、背景干扰以及运动过程，对比总结了基于机器视觉的人体姿态识别方法和基于机器学习的避碰策略，详细比较各类方法的研究现状及应用，并探讨了基于深度学习的目标识别和避碰方法的发展及应用。

The integration of intelligent robots and human intelligence, that is, human-machine collaborative integration, has realized that the mechanical advantages of robots and the advanced cognitive abilities of humans are concentrated in the same working framework, and can work cooperatively in complex environments, thereby improving efficiency. Aiming at complex human-machine interaction scenarios, especially robots in light condition changes, background interference and motion processes, human posture recognition methods based on machine vision and collision avoidance strategies based on machine learning were compared and summarized, the research status and application of various methods were compared in detail, and the development and application of object recognition and collision avoidance methods were discussed based on deep learning.