基于激光位移传感器的植保无人机避障技术研究

发布时间：2018-12-24 11:59

【摘要】：相比于传统的人工施药和地面机械施药,无人机植保在便利性、安全性、喷洒效率、节水节药等方面都具有更显著的效果。近年来,国内植保无人机以遥控式多旋翼机型发展最快,这种机型更适合在无障碍物的空旷地带进行作业;而我国大片的植保作业区域上却较多地分布着电线杆、树木、电力塔甚至房屋等影响无人机飞行安全的障碍物,人工遥控式的半自动机型难以满足这些区域上的植保作业需求。国内外相关企业和科研机构都在大力研发更加智能化的植保无人机。自动避障作为植保无人机全自动机型的关键技术之一具有重要的研究与应用价值。本文的主要工作有:(1)基于激光位移传感器技术,为植保无人机提出一种新的避障检测方法。该方法包括数据块提取、障碍物基本参数计算以及障碍物模式识别三个部分。提出了根据距离值的有效性,从数据序列中提取障碍物所对应的数据块,以平均角度、平均距离和宽度为基本参数来描述障碍物,以及基于障碍物的宽度、数据块的最大间隙、跳变次数和方差为特征的模式识别分类器的设计。最后为激光位移传感器设计了一套在FLYING-BOX机型上的安装方案,并利用激光位移传感器、PICO-CV01工控机和锂电池设计了避障系统的数据采集模块。(2)完成了避障系统的软件设计。将避障软件划分为三个部分即检测部分、动作部分和异常处理部分。检测部分包括飞前自检、数据采集、数据处理和模式识别。提出了根据俯仰角来修正检测距离,根据偏航角来修正检测角度的数据修正方法。在动作部分为无人机设计了一套避障策略并解释了避障过程中动作修正的必要性,随后详细分析了影响指令生成的因素和指令产生的过程。异常处理功能分布在检测部分和动作部分中,负责监测整个避障系统中各个子功能的运行状态,一旦某个或多个地方出错立刻生成相应的错误编码并上传上位机。上位机根据异常情况的不同,采取不同的反应。(3)针对避障检测的准确性和避障系统的有效性进行了实验验证。第一个实验验证避障系统对障碍物角度的检测准确性,第二个实验验证避障系统对于障碍物距离的检测准确性,第三个实验验证避障系统对于树木这种类型障碍物的检测准确性,包括角度值、距离值的检测准确性以及模式识别的准确性,第四个实验,是室外飞行测试,验证无人机是否能够按照事先设计的避障策略完成对障碍物的动态绕飞。实验表明,本文所提的基于激光位移传感器的避障系统能够有效地检测出未知环境下障碍物的角度和距离,能够对植保作业环境中典型障碍物做出较为准确的类型判别,并实现了飞行过程中的动态避障,验证了本文基于激光位移传感器的避障系统的有效性。
[Abstract]:Compared with traditional artificial and ground mechanical application, UAV plant protection has more remarkable effects in convenience, safety, spraying efficiency, water saving and pesticide saving. In recent years, the remote-controlled multi-rotorcraft is the fastest developing type of domestic plant protection UAV, which is more suitable to operate in the open area without obstacles. However, in a large area of plant protection in China, there are many obstacles that affect the flight safety of UAVs, such as utility poles, trees, power towers and even houses. The semi-automatic type of artificial remote control is difficult to meet the requirements of plant protection in these areas. Domestic and foreign related enterprises and scientific research institutions are vigorously developing more intelligent plant protection drones. Automatic obstacle avoidance as one of the key technologies of automatic plant protection UAV has important research and application value. The main work of this paper is as follows: (1) based on the laser displacement sensor technology, a new obstacle avoidance detection method for plant protection UAV is proposed. The method includes data block extraction, obstacle basic parameter calculation and obstacle pattern recognition. According to the validity of the distance value, the block corresponding to the obstacle is extracted from the data sequence. The average angle, the average distance and the width are taken as the basic parameters to describe the obstacle, and the maximum gap of the block based on the width of the obstacle. Design of pattern recognition classifier characterized by jump frequency and variance. Finally, a set of installation scheme for the laser displacement sensor is designed on the FLYING-BOX model, and the laser displacement sensor is used. The data acquisition module of obstacle avoidance system is designed by PICO-CV01 industrial computer and lithium battery. (2) the software design of obstacle avoidance system is completed. The obstacle avoidance software is divided into three parts: detection part, action part and exception handling part. The detection part includes pre-flight self-test, data acquisition, data processing and pattern recognition. A data correction method based on pitch angle and yaw angle is proposed. In the part of action, a set of obstacle avoidance strategy is designed for UAV, and the necessity of action correction in obstacle avoidance is explained. Then, the factors that affect instruction generation and the process of instruction generation are analyzed in detail. The exception handling function is distributed in the detection part and the action section, which is responsible for monitoring the running status of each sub-function in the whole obstacle avoidance system. Once one or more places go wrong, the corresponding error codes are generated and uploaded to the host computer. The host computer adopts different reactions according to the abnormal situation. (3) the accuracy of obstacle avoidance detection and the effectiveness of obstacle avoidance system are verified experimentally. The first experiment verifies the accuracy of obstacle avoidance system for obstacle angle detection, and the second experiment verifies the accuracy of obstacle avoidance system for obstacle distance detection. The third experiment verifies the accuracy of obstacle avoidance system for tree obstacle detection, including angle value, distance value and pattern recognition accuracy. The fourth experiment is outdoor flight test. Verify that UAV can complete the dynamic flight of obstacles according to the pre-designed obstacle avoidance strategy. The experimental results show that the obstacle avoidance system based on laser displacement sensor can effectively detect the angle and distance of obstacles in unknown environment, and can make a more accurate classification of typical obstacles in plant protection environment. The dynamic obstacle avoidance in flight is realized, and the effectiveness of the obstacle avoidance system based on laser displacement sensor is verified.
【学位授予单位】：杭州电子科技大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：V279;V249;TP212

【参考文献】