无人驾驶农机自主作业路径规划方法

针对无人驾驶农机自主作业的应用需求,该研究设计了一种基于区块套行作业模式的路径规划方法,以生成含有速度指令和机具状态指令的可执行路径,重点解决田内作业的四边形地块适应性、无人驾驶农机适应性和农田作业路径完整规划等问题。该方法由农田信息处理模块和路径规划模块组成,农田信息处理模块将测绘产生的地块轮廓数据和障碍物数据处理为便于运算的地块轮廓点数据和障碍物轮廓点数据形式,然后由路径规划模块利用用户输入的作业方向、作业幅宽、转弯半径和起始方位等作业参数,经过作业梯形区生成、掉头区与作业区划分、作业条带分割、障碍物条带处理、作业条带路由、掉头路径生成和最终指令路径生成等子模块,最终生成无人驾驶农机的指令路径。仿真试验结果表明,相对于相邻法,该方法的作业面积比及作业路程比分别提升了10.0%和8.8%。播种作业田间试验结果表明,无人驾驶农机自主作业的横向偏差的均值和标准差分别为左偏0.002和0.027m,满足作业要求。研究结果表明,该研究提出的方法适应不同的四边形农田和障碍物,可以结合不同的作业参数完成路径规划,能够满足无人驾驶农机自主作业的需求。

Autonomous operation path planning method for unmanned agricultural machinery

Abstract: This study aims to meet the application requirements of unmanned driving and autonomous operation in agricultural machinery, particularly for better adaptability of quadrilateral farmland. A path planning was also designed to implement the farmland operation using unmanned agricultural machinery. AB line operation mode was selected to generate executable paths with speed and implement state instructions. Two parts were composed of farmland information processing and path planning. The farmland information processing module was used to process the surveying point data in land contour and obstacles. Among them, one data processing was for land contour surveying and mapping points, and another for obstacle surveying and mapping points. The former was used to extract the longitude and latitude fields of original position data, and then convert them into Universal Transverse Mercator (UTM) coordinates. The latter was utilized to convert the point and line obstacles generated by surveying and mapping into the unified obstacle polygonal contour data for the subsequent path planning. Next, the operating parameters were defined by the user in the path planning module, including the operating direction, operating width, turning radius, and starting position. The sub-modules were also constructed, such as unmanned operation trapezoidal area generation, turning and operation area division, operation strip segmentation, obstacle strip processing, strip routing planning, turning, and final path generation. The result showed that: (1) In the sub-module of Unmanned operation trapezoidal area generation, the operation direction was taken as the parallel side to construct a trapezoidal area suitable for unmanned operation in the quadrilateral land. (2) In the sub-module of turning and operation area division, the width of the turning area was set, according to the operating width and turning radius. Then, 2 parallelogram turning and 1 operation area were generated inside the unmanned operation trapezoidal areas. (3) In the sub-module of operation strip segmentation, the operating direction was taken as the strip direction, while, the operating width as the strip width, in order to complete the strip dividing of the entire operation areas. (4) In the sub-module of obstacle strip processing, the operation line was cut to intersect with the obstacle polygon back and forth, and thereby to construct a detour path composed of arcs and line segments. (5) In the sub-module of strip routing planning, the operation strips were sorted in order. All the strips were divided into the blocks with constructing units, while, the strip selection was designed for different operating sequences to achieve better orderly operation strips. (6) In the sub-module of turning path generation, a U-shaped turning path was generated in the form of "arc-line-arc" using the adjacent sequence of operation strips and turning radius. (7) In the sub-module of final path generation, the final path was realized by adding speed instructions and implement state instructions at the path points. The final path was thus executed in the unmanned agricultural machine. Simulation tests show that the proposed method was suitable for different quadrilateral farmland and obstacles. The operation area ratio and distance ratio increased by 10.0% and 8.8%, respectively, compared with the adjacent. Field tests show that the mean value and standard deviation for the lateral deviations of agricultural unmanned driving and autonomous operation were 0.002 m to the left and 0.027 m, respectively. Consequently, the generated path can meet the requirements of unmanned and autonomous operation in agricultural machinery. The finding can provide a complete path planning for unmanned agricultural machinery with the high adaptability of farmland and operating parameters for higher production efficiency.