温室超声波驱虫机器人导航系统设计与试验

Design and Experiments of Greenhouse Ultrasonic Pest-control Robot Navigation System

摘要:

为提高温室环境下超声波驱虫的机械化与自动化水平,本文根据温室规范化栽培模式下的高效驱虫需求,设计了超声波驱虫机器人,并根据其自主作业模式设计了该机器人的自主导航系统。该导航系统使用基于点云滤波优化的三维激光雷达Cartographer 3D算法对作业环境建图,并最终形成导航所需的栅格地图,使用自适应蒙特卡罗定位算法进行多传感器融合定位,融合遍历式路径平滑插值优化方法与基于JPS跳点搜索的路径直线优化方法改进Dijkstra全局路径规划算法,并结合TEB局部路径规划算法进行路径规划。实验表明:融合点云滤波后构建的温室环境地图长宽误差分别为0.035、0.019 m;改进Dijkstra算法相比A?与传统Dijkstra算法在直线行间规划路径长度分别减少0.64%与3.12% ,在弯曲场景规划路径长度分别减少0.41%与5.65% ;大于60°折点数量在直线行间分别减少71.4%与75% ,在弯曲行间分别减少20%与36.8% ,且离散情况更优;对机器人进行导航运动实验,其横纵平均偏差分别为0.048m与0.089m,航向平均偏差为8.26°,满足超声波驱虫机器人在温室行间自主作业导航需求。最后进行了超声波驱虫机器人的驱虫效果实验,使用超声波驱虫机器人使斜纹夜蛾产卵数量减少98.92%。

Abstract:

Aiming to enhance the mechanization and automation levels of ultrasonic pest control in greenhouse environments, an ultrasonic pest control robot was designed based on the efficient pest control requirements under standardized greenhouse cultivation models and the autonomous navigation system for the robot was designed according to its autonomous operation mode. The navigation system used the 3D LiDAR Cartographer 3D algorithm optimized based on point cloud filtering to map the operating environment and finally form the raster map required for navigation, using the adaptive Monte Carlo localization algorithm for multi-sensor fusion localization, and fused the traversal path smoothing interpolation optimization method with the path straight line optimization method based on the JPS jumppoint search to improve Dijkstra's global path planning algorithm, and combined with TEB local path planning algorithm for path planning. Experiment results showed that the length and width errors of the greenhouse environment map constructed after fusing point cloud filtering were 0.035m and 0.019m, respectively; the improved Dijkstra algorithm reduced the length of the planned paths between straight lines by 0.64% and 3.12% compared with A? and traditional Dijkstra algorithms, and the lengths of the planned paths in curved scenes were reduced by 0.41% and 5.65% . The number of folds greater than 60 degrees was reduced by 71.4% and 75% in straight rows and 20% and 36.8% in curved rows, and the discretization was better. The robot was subjected to navigation movement experiments, and its average transverse and longitudinal errors were 0.048m and 0.089m, respectively, and the average heading error was 8.26°, which satisfied the autonomous operation and navigation needs of the ultrasonic pest-control robot in the greenhouse rows. Finally, the deworming effect experiment of ultrasonic deworming robot was carried out, and the number of eggs laid by Twillow moth was reduced by 98.92% by using ultrasonic deworming robot.