室外自主搜救飞行机器人的设计与实现

发布时间：2018-01-08 10:34

本文关键词：室外自主搜救飞行机器人的设计与实现　出处：《西南科技大学》2017年硕士论文　论文类型：学位论文

【摘要】：近年来随着户外运动的兴起,缺乏户外常识的游客迷失野外的事件时有发生。在应对野外搜寻及搜救任务方面,我国现有的野外搜救多以地面式人力搜索或通用航空人眼观察等方式进行,在野外复杂环境中搜救效果往往不明显,尤其在夜晚搜救过程中更是受到了较大的限制。本文主要针对野外搜救问题设计制作了一种具有自主搜救功能的飞行机器人(Autonomous Searching Aircraft,ASA)。该飞行机器人采用捷联式惯性导航系统对飞行器姿态进行解算,并融合GPS、气压计以及惯导数据对ASA的位置进行估计,最后通过串级PID控制算法实现对飞行器姿态及位置的控制。同时,该搜救飞行机器人能够根据地面站对其设定的搜救区域进行自主航迹规划,从而实现自主、高效的空中搜救任务。自主搜救飞行机器人通过机载的无人机专用红外热像仪FLIR TAU2对搜救区域进行图像采集,并将拍摄的红外图像传输给机载的高性能嵌入式处理器ODROID-XU4进行实时处理,实现对红外图像中人体目标的实时检测,将检测到的人体目标的地理位置发送给地面站,以便搜救人员能够及时高效地完成救援任务。在红外人体目标检测方面,采用HOG(Histogram of Oriented Gradient,方向梯度直方图)对红外目标进行特征描述,并通过线性SVM(Support Vector Machine,支持向量机)分类器对HOG特征分类检测。同时,本文采用自建训练集对SVM进行了训练以满足ASA对红外人体目标检测的需求。实验表明,本文所设计的算法流程结构简单、有效性强,所设计的ASA能够实现对野外人员的自主搜救任务。
[Abstract]:In recent years, with the rise of outdoor sports, lack of knowledge of the outdoor tourists lost in the wilderness, when the incident occurred. In response to the search and rescue mission field, field rescue in our country existing to more general aviation ground manpower search or eye observation, and the effect is often not obvious in the complex environment, especially in the night search and rescue process is restricted. This article mainly aims at the problem of field rescue in the design of a flying robot with independent search function (Autonomous Searching Aircraft, ASA). The flying robot using strapdown inertial navigation system on vehicle attitude is solved, and the fusion of GPS and inertial position, pressure gauge the data of the ASA estimation, finally through the cascade PID control algorithm to control the aircraft attitude and position. At the same time, the search and rescue robot can fly According to the ground station for autonomous path planning of the search and rescue region set, so as to realize autonomous aerial search and rescue mission. The rescue robot autonomous flight, image acquisition of UAV airborne search and rescue region by special infrared thermal imager FLIR TAU2, and the transmission of infrared image taken for airborne high performance embedded real-time ODROID-XU4 processor processing, to achieve real-time detection of human targets in infrared image, geographical location to send a human target will be detected to the ground station, in order to search and rescue personnel to complete the rescue mission timely and efficiently. In the infrared human target detection, using HOG (Histogram of Oriented Gradient, the direction of the gradient histogram) for feature description of infrared target. According to the linear SVM (Support Vector Machine, a support vector machine classifier) detection of HOG feature classification. At the same time, this paper uses self The training set is trained for SVM to meet the needs of ASA for infrared human body detection. Experiments show that the algorithm is simple and effective, and the designed ASA can achieve autonomous search and rescue tasks for field personnel.

【学位授予单位】：西南科技大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.41;TP242

【相似文献】