Precise target characteristics detection could provide important parameters for smart variable-rate sprayers. In order to diminish the influence of complex terrain conditions, 3 kinds of dynamic attitude angle deviation correction methods were proposed to mitigate the errors of Lidar-based spray target detection caused by roll angle, pitch angle and yaw angle respectively. An experimental vehicle integrated with a laser scanning sensor detection unit and an inertial measurement unit (IMU) was used to detect spray targets under complex road conditions. The laser sensor detection unit was capable of detecting spray targets with the laser scanner. The inertial measurement unit was able to detect the real-time attitude angle deviations of the vehicle. The dynamic roll correction adopted real-time measured roll angles to correct detection targets and 3-D (three-dimensional) reconstruction images using re-matching trigonometric function and laser scanning polar coordinate value. For the dynamic pitch correction, re-combination of laser scanning target frames and detection points under the guidance of pitch angle values averaged segmentally was used to diminish dynamic pitch angle deviations. The coefficient value of laser scanning depth data combined with yaw angle average values was applied for the dynamic yaw correction. According to the measured vehicle dynamic attitude angle deviations, these 3 kinds of dynamic attitude angle deviation correction methods were put forward to obtain accurate characteristics of the targets and 3-D reconstruction images. The verification experiments for the proposed correction methods to overcome complex field road conditions were divided into 3 test steps. Firstly, an artificial tree and a cuboid chosen as targets were detected with specified detection distances and laser travel speeds to verify pitch correction algorithm when single dynamic pitch angle changed under downhill or uphill terrain. Secondly, the artificial tree was selected as the laser scanning target to test the correction algorithms when dynamic pitch angles combined with roll angles existed under long wooden unilateral bridge terrain. Finally, the artificial tree and the cuboid cabinet were selected as the laser scanning target to verify the correction methods under uneven complex terrain. The data correction process and 3-D image reconstruction were conducted using MATLAB software. The experiment results of 3 steps showed that the maximum relative errors of the height and width of cuboid cabinet were 8.89% and 8.00% respectively after the correction. The relative errors of the height, width and canopy height of the artificial tree were 5.63%, 10.00% and 5.00%, respectively. The 3-D reconstruction images also had significant improvements after the correction. The test results verify the effectiveness of the proposed data correction methods for laser attitude angle deviations correction under complex road conditions.%基于高精度激光传感器的喷雾靶标特征检测是精准施药变量决策的重要依据.为了改善复杂地形条件对车载激光靶标检测的影响,进行了车载激光喷雾靶标检测与矫正研究.该文基于惯性测量单元(inertial measurement unit,IMU)与UTM-30LX型激光传感器搭建靶标检测试验车,IMU实时获取车体姿态角的偏航角、俯仰角及侧倾角信息,车载激光传感器实时获取目标切面轮廓的极坐标数据.将获取的目标切面轮廓的极坐标数据与试验车姿态角信息相匹配,通过矫正算法获取精确的目标外形尺寸信息并重构目标三维图像.试验设计首先对长方体柜子与仿真树进行车体单一动态俯仰角的检测试验,然后以仿真树为试验目标,进行车体存在复合动态俯仰角与侧倾角的检测试验,最后在未知地形条件下对长方体柜子以及仿真树进行动态姿态角检测与矫正试验.利用MATLAB软件对数据矫正分析,对矫正后的目标尺寸信息进行误差分析并重构目标三维图像.试验结果显示矫正后长方体柜子的高度、宽度最大相对误差分别为8.89%和8.00%,仿真树的高度、宽度以及树冠高度最大相对误差分别为5.63%、10.00%和5.00%,矫正效果良好,验证了矫正算法的有效性.
展开▼
