树冠环绕式仿形对靶施药机设计与试验

    Design and test of the canopy wrap-around profiling-to-target sprayer for orchards

    • 摘要: 针对树冠横切面形状引起施药机喷头与树冠间距发生动态变化从而导致树冠内、外和两侧、中部等区域喷药量不均匀、较多药液喷洒在无效空间区域的问题,该研究提出一种基于激光雷达树干定位的树冠环绕式仿形对靶喷药方式。首先,基于喷头环绕树冠的运动需求,设计对称交叉布置的两自由度仿形机构,并建立喷头运动学模型,采用基于二环PID算法对喷杆伸缩和喷头旋转进行精准控制,通过匹配施药机前进速度实现喷头半圆和多边形轨迹仿形控制;然后采用平面激光雷达对树干高度的水平面进行扫描,提出基于DBSCAN(dnsity-based spatial clustering of applications with noise)密度聚类和3点树干形状拟合的树干动态识别和定位方法;再次,在FreeRTOS框架下搭建双层控制系统,对树干定位感知、动态喷头仿形轨迹控制、数据通信和操控交互等多任务进行并行处理,实现基于激光雷达树干位置信息的喷头动态伺服仿形控制。最后,以雾滴沉积量、雾滴密度、药液覆盖率和雾滴体积中值直径作为量化指标,在树冠不同区域布置10个检测点,对环绕式和定距施药开展6组对比试验。雾滴结果表明,树干纵向和横向定位误差分别为9.44和1.74 cm,环绕式仿形施药方式的平均雾滴密度为72.2 个/cm2,平均沉积量为1.99 μL/cm2,药液覆盖率为47.5%,相比定距施药方式,雾滴沉积量和雾滴密度分别提升36.3 %和58.3 %,雾滴沉积量变异系数降低了60%,环绕式仿形对靶喷雾可有效提高药液利用率和喷药均匀性。

       

      Abstract: Precision spraying technologies have been widely applied in modern agriculture, such as profiling, targeting, and variable spraying. But there are still great challenges, in terms of uneven distribution of canopy sap, low sap utilization, high dosage of sap, and serious environmental pollution. Especially, the uneven spraying volume can be found in the cross-sectional shape of the tree canopy caused by the dynamic changes in the spacing between the nozzle and the canopy. It is a high demand for a high effective space area in the spraying volume of liquid. In this study, a novel canopy wrap-around profiling-to-target spraying was proposed to draw on the manual spraying actions, particularly with the nozzle completing dynamic surround profiling of each fruit tree during the advance of the sprayer and targeting the inside of the canopy for spraying. A spraying machine of wind-fed fruit trees consisted of five subsystems: the power platform, profiling-to-target mechanism, wind-fed spraying system, navigation control system, and canopy detection system. There were also environment perception, automatic navigation, and profiling-to-targeting functions. A symmetrical mechanism was also designed with cross-arranged, two-degree-of-freedom, and radial profiling, according to the nozzle motion requirements. The forward motion of the sprayer was combined with the position and attitude adjustment of the nozzle in plane space. Besides, the kinematic model of the nozzles was established to accurately control the lateral and rotational movement position and speed using the two-loop PID controllers. Then, the semicircular and polygon trajectory profiling control of the nozzle was achieved to match the motion speed of the sprayer. At the same time, the nozzle was rotated dynamically to keep the spray aligned with the center of the canopy, in order to reduce the drift of the liquid in the air. After that, trunk identification and localization were proposed using DBSCAN density clustering and 3-point trunk shape fitting, where the planar LIDAR scanned the horizontal plane of trunk height in real time during the operation of the sprayer. The 2-D LiDAR was employed and mounted horizontally on the front of the applicator to scan and identify the tree trunk in advance. The localization of the nearest trunk to the sprayer was sent to the control system in real time, in order to enable the trunk position tracking for the wrap-around profiling control. The results showed that the positioning errors of the longitudinal and lateral trunks were 9.44 and 1.74 cm, respectively. The total control system of the applicator was equipped with an STM32 processor using the FreeRTOS real-time operating system. The dynamic servo profiling control of the nozzle was realized using LiDAR trunk position information via the multi-task distribution and efficient processing of real-time trunk position information sensing, dynamic nozzle profiling trajectory control, data communication, and operation interaction. Finally, the quantitative indexes were taken as the droplet deposition, droplet density, droplet coverage, and median diameter of droplet volume. 10 test points were also arranged in the different areas of the canopy. Furthermore, the six sets of comparison tests were carried out between the wrap-around profiling and fixed-spacing spraying mode. It was found that the average droplet density of the wrap-around spraying mode was 72.2 droplets/cm2, the average deposition volume was 1.99 μL/cm2, and the droplet coverage rate was 47.5%. More importantly, the droplet deposition and density increased by 36.3%, and 58.3 %, respectively, compared with the fixed-spacing spraying mode. Therefore, a higher liquid utilization rate was achieved during this time. The coefficient of variation of droplet deposition was 27.7% for the wrap-around application, which was 60% lower than that in the fixed-spacing spraying mode. There was a significant improvement in the canopy application uniformity. In summary, the wrap-around profiling spraying mode can be expected to effectively improve the utilization rate and uniformity of precision spraying.

       

    /

    返回文章
    返回