基于深度双目视觉处理的智能采摘机器人设计

Design of Intelligent Picking Robot Based on Deep Binocular Vision Processing

针对采摘机器人对果蔬的位置定位不够准确、无法准确避障,导致采摘效率较低的问题基于深度双目视觉处理对智能采摘机器人进行了设计。智能采摘机器人的主要组成包括PLC控制器、视觉系统、移动平台、导航系统、机械臂、通信系统和电源。为了对采摘机器人的机械臂进行最优路径规划并避障,通过对采集的图像进行预处理后,利用双目视觉系统对果蔬进行精准定位,然后采用哈夫变换直线检测的方法进行最优路径的设计和选择,最终确定最优采摘路径。对采摘机器人进行运动轨迹精度试验和采摘试验,结果表明:采摘机器人对果蔬的采摘成功率较高,可以满足果农对于采摘机器人的要求。

Aiming at the problem that picking robot’s location of fruits and vegetables was not accurate enough and it could not avoid obstacles accurately,which leaded to low picking efficiency,the intelligent picking robot based on deep binocular vision processing was designed.The intelligent picking robot was constituted of PLC controller,vision system,mobile platform,navigation system,mechanical arm,communication system and power supply.To carry out optimal path planning and obstacle avoidance for the mechanical arm of the picking robot,the binocular vision system was used to ac-curately locate fruits and vegetables after the pre-processing of the collected images,and then the method of Huff trans-form linear detection was used to design and select the optimal path,and finally determine the optimal picking path.The picking robot movement trajectory accuracy test and picking test were done,the test results show that the picking robot for fruit and vegetable picking success rate is high,and it could meet the requirement of fruit farmers for picking robot.