音频感知哈希闭环检测的无人机仿生声呐SLAM算法研究

PDF下载HTML

Title:

Research on BATSLAM algorithm for UAV based on audio perceptual hash closed-loop detection

作者:

胡文超, 孙新柱, 陈孟元

安徽工程大学 安徽省电气传动与控制重点实验室, 安徽 芜湖 241000

Author(s):

HU Wenchao, SUN Xinzhu, CHEN Mengyuan

Anhui Key Laboratory of Electric Drive and Control, Anhui Polytechnic University, Wuhu 241000, China

关键词:

同步定位与地图构建; 音频感知哈希; 无人机; 闭环检测; 仿生声呐系统; 经历图; 绝对差值和; 耳蜗图

Keywords:

simultaneous localization and mapping; audio perceptual hash; unmanned aerial vehicle; closed-loop detection; Experience diagram; bionic sonar; sum of absolute difference; Cochlear map

分类号:

Q811.211;TP751

DOI:

10.11992/tis.201708018

摘要:

针对基于SLAM技术无人机在特定高度下构建二维经历图的优化问题，在RatSLAM的基础上，采用仿生声呐系统代替视觉传感器的BatSLAM模型和音频感知哈希闭环检测，实现在暗光条件下的二维经历图优化。BatSLAM模型通过绝对差值和（SAD）图像处理方法来进行仿生声纳模板的更新，此方法仅仅判断二幅耳蜗图外观是否一致，不存在几何处理和特征提取。由于耳蜗图在获取和传输过程中会产生各类噪音，相同位置获得的耳蜗图具有一定的差异，会导致构建的经历图失真。本文在BatSLAM的基础上，使用音频感知哈希算法对耳蜗图进行特征提取，并进行闭环检测。改进后的算法不仅考虑到外观，而且考虑到相邻频带间的能量差异，通过提高闭环检测准确率，来改善经历图的失真问题。仿真实验表明：采用基于音频感知哈希闭环检测的BatSLAM模型，不仅实现了无人机特定高度和暗光条件下二维经历图的构建，而且提高了闭环检测准确率，从而改善经历图的失真问题，实现经历图的优化。

Abstract:

Aiming at the problem of constructing two-dimensional empirical graph based on SLAM technology at certain heights,on the basis of RatSLAM, using the bionic sonar system to replace the visual sensor’s BatSLAM mode and audio perceptual hash closed-loop detection, realizing two-dimentional empirical graph optimization under dark conditions. The BatSLAM model uses Sum of Absolute Difference (SAD) image processing methods to update the bionic sonar template. This method only judges whether the appearance of the two cochlear images is consistent, and does not have geometric processing and feature extraction. Because the cochlear images produce various noises during the acquisition and transmission, There are some differences in cochlear maps obtained at the same position. which can lead to the distortion of the constructed empirical map. This article is based on BatSLAM, The audio perceptual hash algorithm is used to extract features of cochlea,and according to closed-loop detection. The improved algorithm not only considers the appearance, but also considers the energy difference between adjacent bands. By improving the accuracy of closed-loop detection, to improve the experience map of the distortion problem. Simulation experiment shows,the BatSLAM model based on audio perceptual hash closed-loop detection is adopted, it not only realizes the construction of two-dimensional experience map under certain height and dark conditions of UAV, but also improves the accuracy of the closed-loop detection, improving the experience map of the distortion problem and implementing the optimization of experience graph.