A review of key technologies on flower picking robot: from perception, planning to non-destructive operations
Flower picking is a labor intensive process heavily in the floriculture industry, and it remains predominantly manual. With the increasing shortage of agricultural labor and the continuous rise in labor costs, the sustainable development of the flower industry is facing severe challenges. Therefore, the using of robotic systems to picking flower plants has become an inevitable trend in the development of modern agriculture. This paper presents a systematic review of recent advances in flower picking robot, with a particular focus on system architecture and core enabling technologies. The core technologies are visual perception and target recognition, motion control and path planning, and flexible, non-destructive end effector design. The research status and limitations of these key technologies are analyzed, with special attention to their applications in both open field and controlled facility environments. Current challenges include low recognition accuracy under complex lighting and occlusion, insufficient end effector adaptability to diverse varieties, and low overall picking efficiency. Future directions include multiple modal sensing integrated with artificial intelligence, bionic and soft robotic end effectors, and multiple robot collaborative system. Finally, based on the perspectives of design and application, the review provides future research directions toward industrialization, with a particular focus on key enabling technologies of these technologies.