Fruit and vegetable picking is an important link in China’s agricultural production. The existing manual picking method has high labor intensity and low efficiency. The application of picking robots can greatly improve the efficiency of picking operations and reduce the dependence on manpower. Its research has great practical value. The machine vision system of the picking robot is currently a hot and key research area in this field. This article analyzes the key issues faced by the machine vision system research in terms of precise identification of picking objects, spatial positioning, and path planning in complex environments, and classifies and summarizes the latest research results of machine vision at home and abroad.