融合仿形飞行与能效优化的丘陵山地植保无人机导航方法研究

The terrain of hilly region is always complex and volatile, and the low mechanization level has become the bottleneck which restricts the development of agriculture. Rotor unmanned aerial vehicle (UAV) will become an important approach for hilly plant protection, as it is not constrainted by terrain when working. How to realize the reconciliation flight that is applicable to complex terrain becomes the difficulty of precision operation, with three-dimensional space path planning being the key issue which needs to be solved urgently. Aiming at the problem of poor uniformity of spraying in nonplanar operation, in this proposal, we will study the rapid identification and three-dimensional positioning model based on binocular vision, and put forward the autonomous profiling control method for hilly region. The energy consumption for different working path differs significantly because of the terrain of hilly region, information fusing of the terrain information and UAV tension efficiency is utilized to improve the work efficiency, and the hilly region path planning method is explored based on genetic ant colony algorithm. The profiling flight and path planning method is utilized to build UAV navigation flying platform for hilly region plant protection based on multiple information fusion method. The verification and optimization will be done with winter wheat as research object in Longdong area. The research results are of great significance to explore the theory of UAV’s autonomous control in complex environment, and provide the basic model and method to support the high efficiency of plant protection operation in hilly region. And furthermore, it provides feasible ideas for promoting the mechanized application of hilly region and the development of modern agriculture.

丘陵山地地形复杂多变，机械化程度低已成为制约农业发展的瓶颈，旋翼无人机飞行不受地形限制,在丘陵山地植保作业中具有广泛应用前景，如何实现与复杂地形相适应的仿形飞行是精准作业的难点，三维空间路径规划是其亟需解决的关键问题。针对非平面作业时喷施均匀性差的问题，研究构建基于双目视觉的作业目标快速识别及三维定位模型，提出无人机丘陵山地自主仿形控制方法；丘陵山地地形起伏导致不同作业路径能耗差异显著，融合地形信息与无人机力效特性，以优化作业能效为目标，探索基于遗传蚁群算法的丘陵山地作业路径规划方法；融合仿形飞行与路径规划方法，构建基于多元信息融合的丘陵山地植保无人机导航试验平台，以陇东地区冬小麦为对象开展模型与方法的验证优化。研究成果对探索复杂环境下无人机自主控制理论具有重要意义，为保障丘陵山地无人机植保高效作业提供基础模型与方法支撑，对推动丘陵山地机械化应用和现代农业发展提供可行思路。

丘陵山地地形复杂多变，机械化程度低已成为制约农业发展的瓶颈，旋翼无人机飞行不受地形限制，在丘陵山地植保作业中具有广泛应用前景，如何实现与复杂地形相适应的仿形飞行是精准作业的难点。.项目提出了基于双目视觉的无人机丘陵山地自主仿形控制方法，分析了双目图像匹配模型，设计了改进两步标定法对相机进行标定，标定误差均值为0.3489。选用BM算法实现了双目视觉的立体匹配和空间定位算法；融合地形信息与无人机力效特性，以优化作业能效为目标，针对无人机作业时的三维运动特点和自身能量消耗情况，提出了基于模拟退火算法的最优能耗作业路径规划；航迹控制方面，提出了基于GNSS-视觉组合的无人机山地果园植保作业航迹控制方法。分析水平航迹控制中的航向调整过程，开展控制系统整体设计，提出视觉导航实现行内作业航迹调整控制，GNSS导航实现作业行间转场航迹控制流程，选用RGB空间下基于RGB分量线性组合的作物行提取方法，对分割后的作物行进行二次曲线拟合得到行趋势线，计算得到偏航角值，由PID控制器控制实现航向调整。.以陇东地区地形为试验对象，对成果进行了整体测试，测试结果表明，无人机仿形飞行控制中，相对高度突变时，有明显调整过程，当变化较缓时，无人机相对飞行高度变化较小，仿形效果好，无人机稳定飞行的平均误差为0.01m，最大误差0.15m；能效约束最优路径规划方面，在负载实时变化情况下，最优能耗路径比最短路径可节约能耗11.72%，比常规路径可节约能耗32.04%，表明模拟退火算法可在能耗约束条件下对无人机作业的最优能耗路径进行规划，且具有较好的规划效果；航迹控制中，GNSS模块静态定位误差小于0.26m，动态测试最大误差为0.82m，平均误差为0.53m；视觉导航控制的航迹误差为-27~+48cm，平均误差为23cm。.项目成果可满足丘陵山地植保作业要求，且对探索复杂环境下无人机自主控制理论具有重要意义，为保障丘陵山地无人机植保高效作业提供基础模型与方法支撑，对推动丘陵山地机械化应用和现代农业发展提供可行思路。