关键词: 草莓采摘机器人, 路径规划, 改进A*算法, R-FCN目标检测网络

Abstract: In order to realize the mechanization and automation of strawberry picking in greenhouse, a strawberry picking robot used in solar greenhouse was designed and manufactured. The robot can realize autonomous path planning, identify mature strawberries and complete picking during walking. This design uses ROS distributed computing system as the main control network, uses laser radar to build and locate the map of mobile robot, binocular depth camera to identify and locate mature strawberries, and a 6-DOF mechanical arm equipped with flexible bionic grippers to grasp and place target strawberries. This paper also designs a robot software platform， using improved A* algorithm to achieve autonomous path planning and navigation obstacle avoidance; The R-FCN target detection network and binocular vision technology were used to detect and locate mature strawberries. The experimental results show that: the strawberry picking robot can achieve target detection and positioning, the error between the detected strawberry coordinates and the robots paw coordinates is less than 4 mm, and the recognition rate of mature strawberries is 95%, meeting the picking requirements.

Key words: strawberry picking robot, path planning, improved A* algorithm, R-FCN target detection network