基于视觉检测的苹果采摘机器人系统设计与实现

摘要

In order to improve the automation and intelligence level of apple picking,reduce the labor intensity of human and the damage rate of the fruit, the paper developed a wheeled robot system for automatic detection and picking ripe ap-ples.The system is made up of hardware platforms and software platform.The hardware platform includes four-wheel drive off-road car, IPC-610L, image data ac-quisition card, four degrees of freedom manipulator and end-effectors;The software platform includes Visual C ++6.0 development environment, based on binocular vision and image process-ing techniques to achieve the identification and location of apple, then through the manipulator path planning to achieve the apple picking.Through simulation and data analysis show that the robot can realize automatic n-avigation, automatic recognition, automatic picking apples and other functions in the case of unattended.The correct rate of apple recog-ni-tion and the success rate of picking was 94.00%and 91.33%, respectively.The average period of picking is about 11s and the accuracy and stability of the system is excellent.%为提高苹果采摘的自动化与智能化水平，降低重复繁琐的人工劳动强度，减少对果实的损坏率，研制了一款用于苹果成熟自动检测并采摘的轮式机器人系统。系统由硬件平台和软件平台两部分组成。其中，硬件平台由四轮驱动越野小车、IPC－610 L工控机、图像数据采集卡、四自由度机械臂和末端执行器组成；软件平台基于VisualC＋＋6．0开发环境，使用双目立体视觉技术和图像处理技术实现对苹果的识别与定位，再通过机械臂的路径规划实现对苹果的采摘。通过仿真实验和数据分析表明：机器人在无人值守的情况下，能实现自动导航、自动识别、自动采摘苹果等功能，并且识别成功率大于94．00％，采摘成功率达到91．33％，平均采摘周期约为11s，具有较高的准确性及稳定性。