项目名称: 四旋翼飞行器基于视觉的目标跟踪及自主动态降落
项目编号: No.61305128
项目类型: 青年科学基金项目
立项/批准年度: 2014
项目学科: 自动化技术、计算机技术
项目作者: 刘一莎
作者单位: 大连海事大学
项目金额: 24万元
中文摘要: 以小型四旋翼飞行器基于视觉的运动目标跟踪及面向地面移动平台的自主动态降落为研究对象,针对混合粒子滤波算法中的信息交互问题,以及四旋翼飞行器与地面运动目标间的相对速度与相对位姿估计等问题,通过对不同滤波算法的定性对比分析,从具有鲁棒性的视觉目标跟踪算法研究入手,重点研究基于光流与惯导数据融合的飞行器相对速度估计问题,并基于对地面移动平台顶端着陆标志的实时观测,给出可满足相对速度与相对位姿约束并能自主调整降落轨迹的动态降落新方法,并给出基于实际四旋翼飞行器平台的具体实现。本项目的意义在于解决四旋翼飞行器从手动遥控到全自主进行目标跟踪与自主动态降落等任务时所出现的新问题,并提出合理、实用的技术和方法,从而为该方法的实用化提供可靠的技术支持和原型验证。
中文关键词: 四旋翼飞行器;单目视觉;飞行控制;目标检测与跟踪;场景理解
英文摘要: In this project we will investigate the problems of vision-based moving object tracking and autonomous landing of a quadrotor UAV on mobile platform. Considering the difficulty of information fusion in hybrid particle filter algorithm and the relative velocity/pose estimation between the quadrotor UAV and mobile ground platform, we adopt a technique of comparative analysis on different filtering algorithms and start with robust dynamic target tracking approach. Furthermore, we lay emphasis on the relative velocity estimation based on the fusion between optical flow and IMU data. The key problem in our project is to present a novel autonomous landing approach for a quadrotor UAV on mobile platform which is satisfied with the constraints of relative velocity/pose while the quadrotor UAV can extract the landing mark feature on the top of mobile robot. Throughout this project we will investigate the new problems stemming from autonomous target tracking and landing instead of past remote manual control mode. The final goal of this project is the development of an integrated technique for object tracking and autonomous landing of a Quadrotor UAV on moving platforms as well as providing technical support for future practical application in real Quadrotor UAVs.
英文关键词: Quadrotor UAV;Monocular Vision;Flight Control;Target Detection and Tracking;Scene Understanding