摘要
闭环探测效率不高、视觉节点冗余度大制约着移动机器人视觉导航系统的性能。为了解决这个问题,从两个方面对视觉SLAM算法的关键环节进行了改进:在机器人导航的闭环探测环节采用了一种新的场景相似性测量方法,有效地提高了闭环探测的效率;在视觉节点的生成环节,算法采用了场景之间共有信息量减少的减量式节点探测方法,有效地降低了地图节点的冗余度。仿真和移动机器人实验对方法的有效性和实时性进行了验证,实验结果表明,移动机器人在视觉导航过程中,闭环探测的有效性达到99%以上,平均计算时间为0.03s,地图节点冗余度为0,使得导航系统在闭环探测和构建的地图质量两个方面的性能得到了进一步的提升。
Both low efficiency of loop closure detection and high redundancy of visual nodes in environmental map negatively affect the performance of mobile robot visual navigation.For addressing the problem,the method of visual navigation is improved in terms of loop closure detection and key frame detection: a novel similarity measurement for scenes is adopted to detect loop closure that improving the efficiency of loop closure detection and the method based on mutual information of images is adopted to select key frames as visual nodes that reducing effectively redundancy of nodes.Effectiveness and efficiency of proposed method are verified by simulation and experiment on mobile robot in outdoor environments,the result shows that in visual navigation,the efficiency of loop closure detection is up to 99%,the average calculating time is 0.03s(33 fps),the redundancy is 0,so the performance of robot navigation in terms of loop closing detection and quality of map is improved further.
出处
《计算机工程与设计》
CSCD
北大核心
2013年第8期2892-2896,共5页
Computer Engineering and Design
基金
高等学校博士学科点专项科研基金项目(200805611091)
海南省琼海市公安局委托基金项目(2011H012)
关键词
视觉导航
场景相似性
闭环探测
视觉节点
地图构建
visual navigation
similarity measurement
loop closure detection
visual node
mapping