中国农业机械化科学研究院集团有限公司 主管

北京卓众出版有限公司 主办

周达.农机收获作业参数化路径生成方法[J].农业工程,2022,12(12):13-21. DOI: 10.19998/j.cnki.2095-1795.2022.12.003
引用本文: 周达.农机收获作业参数化路径生成方法[J].农业工程,2022,12(12):13-21. DOI: 10.19998/j.cnki.2095-1795.2022.12.003
ZHOU Da.Parameterized path planning for reaping operations[J].Agricultural Engineering,2022,12(12):13-21. DOI: 10.19998/j.cnki.2095-1795.2022.12.003
Citation: ZHOU Da.Parameterized path planning for reaping operations[J].Agricultural Engineering,2022,12(12):13-21. DOI: 10.19998/j.cnki.2095-1795.2022.12.003

农机收获作业参数化路径生成方法

Parameterized Path Planning for Reaping Operations

  • 摘要: 为使无人驾驶农机在收获作业时准确收割作物,其导航路径应尽可能与播种作业轨迹保持一致。播种作业轨迹通常可近似为直线,因此传统的AB线模式可以满足大多数作业需求。但在某些作业场景中,受客观条件限制,播种作业路径是弯曲且不规则的。为使农户以其熟悉的打点方式设定路径,提出了一种参数化路径生成方法。在农户设定路径时采集车辆主天线轨迹坐标和姿态,解算出车辆后轴中点坐标。通过算法过滤掉轨迹中聚集点和离群点并提取出能够代表轨迹形状的一系列节点。采用分段插值方法,生成基于3阶贝塞尔曲线的参数化路径。以雷沃M1004型拖拉机为试验平台,在北京市昌平区小王庄试验站采集数据。试验结果表明,在平均作业速度5 km/h、聚集点抽稀阈值1 cm、离群点曲率阈值1 m−1、节点抽稀阈值30 cm时,两组参数化路径与各自原始轨迹的最大偏差分别为0.52和0.40 m,说明此方法可以较为准确地还原播种轨迹并生成收获作业目标导航路径。

     

    Abstract: Path planning is crucial for autonomous reaping process and in particular reaping path should be as identical as possible to seeding trajectory so that autonomous vehicles can reap precisely.In the case of autonomous seeding, planned seeding path can be applied directly to reaping process, yet in the case of non-autonomous seeding in which there is no planned path, it is crucial to generate reaping path close to seeding trajectory.The most common applied path planning is the AB line which is generated by setting terminals of a path.However, in some specific cases, paths may be limited to curved and irregular shapes, hence AB line is no longer applicable.Through research, a lot of Chinese farmers prefer path planning similar to AB line when it comes to cases mentioned above.In order to meet needs presented by farmers, a solution for generating parameterized reaping paths wass proposed.Firstly, coordinates and attitudes of main antenna on a seeding vehicle were sampled and used for calculating coordinates of rear axis midpoint.Next, aggregated points and outlier points were eliminated and path nodes were picked which showed general shape of seeding trajectories.Finally, parameterized paths were generated using piece-wise interpolation based on cubic Bezier curves.LOVOL M1004 tractor was used as experiment platform and data was sampled in experimental station in Xiaowangzhuang, Changping District, Beijing City.Experiment results showed that maximum lateral errors between parameterized curves and raw trajectories were 0.52 and 0.40 m in two experimental groups under condition that average operation speed being 5 km/h, compression threshold for aggregated points being 1 cm, curvature threshold for outlier points being 1 m−1 and compression threshold for path nodes being 30 cm.This solution proposed was confirmed to be effective in precisely restoring seeding trajectories restoration and generating reaping paths.

     

/

返回文章
返回