摘要
采用机器人自动去除樱桃番茄腋芽时,为得到叶柄的位置及水平投影方向角,需调整摄像头位置与角度,以便于采集到主茎、腋芽和叶柄的主平面图像,并使后续机器视觉判别腋芽的步骤得以顺利进行。该文设计了光电式自动旋转测向机构;采用闭合环形机构环绕樱桃番茄主茎,并通过往复旋转方式使固定在活动环上的光电传感器能水平360°扫描叶柄,通过叶柄水平投影角度测量算法计算得到叶柄的水平投影方向角;分析8个光电传感器的运动轨迹,得到其线速度与角速度的关系及扫描系数。经试验表明:扫描系数为1.5,角速度为1.5πrad/s,线速度为20 mm/s时,检测成功率为95%;高度补偿设定为28 mm,成功率为93%;对42株樱桃番茄的153个叶柄进行测试,在1.8 m高度内的成功率为88.2%。研究结果为腋芽生长点判定与摘除提供参考。
In order to remove the axillary bud of cherry tomato automatically by a robot, the camera should be in specific locations to capture images of axillary bud and petiole. Camera optical axis perpendicular to the main plane coexisted with main stem, petiole and axillary bud. The horizontal projection direction angle of petiole in horizontal plane should be measured. In this paper, a photoelectric automatic rotating direction-finding mechanism was designed to get the value of the angle. The end effector of the robot was composed of a direction-finding mechanism, a camera and a pneumatic shear. Telescoping devices were used to control the extension and contraction of direction-finding mechanism. The two half-circle of active circle was closed by mechanical claw, and combined into a complete circle around the main stem. Then active circle moved upward under manipulator control. Meanwhile, the active circle rotated repeatedly at 90 degree angle. The inside diameter of active circle was 60 mm. There were 8 photoelectric sensors uniformly distributed on the active circle, and the included angle between every two sensors was 45°. The photoelectric sensors would be triggered when the petiole appeared 10 mm above the active circle. If the inside diameter of active circle was too big, sensors might well be misguided by leaves on which branches closest to the main stem, and produced great error. On the contrary, if the inside diameter of active circle was too small, active circle might be stuck at the parts of the main stem which deviated from the hang-off line. 8 photoelectric sensors fixed on the active circle were used to detect the existence of petiole, and the angle of petiole in horizontal plane was got through the angle of sensors and the rotation angle of active circle. The values of angles were sent to control module, which moved the end effector to the normal direction of the main plane. The end effector should always be kept in front of the tomato plant. The relations among angular velocity, linear velocity and coefficient were obtained by analysis of active circle movement trajectory. Angular velocity was determined by the linear velocity and the scanning coefficient, which should be set previously, and scanning coefficient must be greater than 0.5. If the value of coefficient was smaller than 0.5, direction-finding mechanism would miss the petiole, and damage the cherry tomato plant. Angular velocity was directly proportional to linear velocity. The greater the linear velocity was, the greater the angular velocity would be, and the greater the errors of horizontal projection direction angle would be. The camera was installed under the direction-finding mechanism. Robot arm drove the end effector up to make the main stem, petiole and axillary bud displayed completely in viewfinder display of camera. This process of height compensation was determined by the height difference between the optical axis of the camera and the active circle. Height compensation made the complete axillary bud displayed on camera lens, so that the robot could locate the growth point of the petiole and then remove the axillary bud. Experimental results showed that detection successful rate was 95% with scanning coefficient as 1.5, angular velocity as 1.5π rad/s, and linear velocity as 20 mm/s. The successful rate was 93% if the height compensation was set as 28 mm. 153 petiole samples had been tested, and the successful rate was 88.2% for the height of cherry tomato plant less than 1.8 m.
出处
《农业工程学报》
EI
CAS
CSCD
北大核心
2017年第S1期42-48,共7页
Transactions of the Chinese Society of Agricultural Engineering
基金
国家自然科学基金面上项目(51375460)
浙江省科技厅公益技术应用研究计划项目(2014C32105)
关键词
传感器
机器人
设计
樱桃番茄
叶柄
方向检测
旋转
水平角度
sensors
robots
design
cherry tomato
petiole
direction finding
rotation
horizontal angle