针对应用于各类移动平台导航的视觉惯性里程计(VIO)在回环检测失败时,系统状态估计存在严重漂移且缺乏全局一致性的问题,在全球导航卫星系统(GNSS)拒止情况下,将室内定位技术超宽带(UWB)代替GNSS加入到视觉惯性里程计可以校正累积误差,提高系统定位精度和稳定性.该文聚焦视觉/惯性/UWB多传感器融合技术,介绍了视觉/惯性/UWB组合导航系统的基本组成,对多传感器时空标定进行了回顾,并梳理了视觉/惯性/UWB数据融合算法,最后对组合系统未来的发展趋势进行了展望.

近年来,弹性PNT和综合PNT不断发展,多源传感器的弹性融合机制受到广泛关注.针对多源组合导航中视觉传感器的测量精度和可靠性受环境影响大,对精度的影响程度难以单独评估等问题,通过研究单目视觉里程计位姿解算质量的评估因素,体现多源传感器融合系统中视觉传感器的测量质量,提出基于李代数推导的反应空间点构型对相机位姿恢复精度影响的精度衰减因子Visual-DOP,设计了单目视觉里程计位姿解算质量评估算法,并在视觉里程计权威数据集KITTI上做了验证,通过控制变量的方法解耦每一个评估因素与定位定姿精度的关系,结果显示采用的位姿解算质量评估因素能够正确反应单目视觉里程计的导航精度受视觉测量质量的影响程度.

在 GNSS 信号拒止的室内环境定位中,针对视觉惯性里程计(VIO)在长期运动或无回环等不利环境下产生误差累积偏移以及超宽带(UWB)受非视距影响定位精度难以保证的问题,提出了一种基于抗差估计的UWB辅助视觉惯性自适应组合定位算法.首先,构建UWB与VIO的优化框架,利用UWB定位结果对VIO进行全局约束.其次,在后端优化阶段加入抗差估计,实时调整传感器间的权重值,抑制UWB非视距的影响.最后,在EuRoc数据集和真实场景中进行了实验验证.真实场景实验结果表明,组合算法在非遮挡条件下定位精度相比于VINS_Mono提高 75.05%,基于抗差估计的组合算法在遮挡条件下定位精度相比于不加抗差估计的组合算法提高 37.53%.

针对旋翼无人机使用合作标志辅助降落过程中,可能出现的误匹配问题,设计了一种融合特征信息和编码信息的合作标志,提出了对应的匹配方法,基于此合作标志的特征信息可以确定尺度,根据编码信息能够排除错误匹配结果.为检验标志识别的效果,搭建了移动式无人机辅助降落平台.无人机自主降落时,使用机载相机捕获地面降落平台上显示的合作标志影像,首先通过标志区域的形状参数进行特征匹配,然后根据编码圆匹配算法解算对应的编码信息,指向对应的降落区域.实验结果表明:通过增加编码信息的约束,能够有效防止错误匹配,单张影像处理过程平均耗时为12 ms,能够满足旋翼无人机自主降落的实时性要求.

To improve localization and pose precision of visual–inertial simultaneous localization and mapping (viSLAM) in complex scenarios, it is necessary to tune the weights of the visual and inertial inputs during sensor fusion. To this end, we propose a resilient viSLAM algorithm based on covariance tuning. During back-end optimization of the viSLAM process, the unit-weight root-mean-square error (RMSE) of the visual reprojection and IMU preintegration in each optimization is computed to construct a covariance tuning function, producing a new covariance matrix. This is used to perform another round of nonlinear optimization, effectively improving pose and localization precision without closed-loop detection. In the validation experiment, our algorithm outperformed the OKVIS, R-VIO, and VINS-Mono open-source viSLAM frameworks in pose and localization precision on the EuRoc dataset, at all difficulty levels.

The sensor drift problem is objective and inevitable, and drift compensation has essential research significance. For long-term drift, we propose a data preprocessing method, which is different from conventional research methods, and a machine learning framework that supports online self-training and data analysis without additional sensor production costs. The data preprocessing method proposed can effectively solve the problems of sign error, decimal point error, and outliers in data samples. The framework, which we call inertial machine learning, takes advantage of the recent inertia of high classification accuracy to extend the reliability of sensors. We establish a reasonable memory and forgetting mechanism for the framework, and the choice of base classifier is not limited. In this paper, we use a support vector machine as the base classifier and use the gas sensor array drift dataset in the UCI machine learning repository for experiments. By analyzing the experimental results, the classification accuracy is greatly improved, the effective time of the sensor array is extended by 4–10 months, and the time of single response and model adjustment is less than 300 ms, which is well in line with the actual application scenarios. The research ideas and results in this paper have a certain reference value for the research in related fields.

针对在全球卫星导航系统(GNSS)及超宽带(UWB)组合的室内外无缝定位系统中,室内定位基准取决于UWB基站的位置基准,与GNSS通常所用的大地坐标系不一致的问题,提出一种基站位置估计方法:利用携带GNSS接收机和UWB标签的移动站来估计室内UWB基站位置;以移动站获取的GNSS坐标和UWB测距信息作为观测量,采用扩展卡尔曼滤波(EKF)法解算UWB基站坐标,使UWB定位基准自然统一到大地坐标系下.实验结果表明:该方法具有较好的估计效果;UWB基站无需提前布设,自然统一到大地坐标系下.

近年来,基于视觉的导航和目标跟踪算法开始广泛应用于无人平台.针对各类视觉导航算法和综合任务方案在无人平台上实地测试风险性高、现场采集参数精度低、效率差等问题,设计并搭建了一套以高精度工控导轨平台、姿态控制云台、工业相机和模拟地形沙盘为主要部件的无人平台半物理仿真系统.引入OptiTrack光学运动捕捉设备对仿真系统真实的位姿控制精度进行了评价,结果显示系统空间三轴位置控制均方根误差在1 mm左右,横滚、俯仰、航向姿态角控制均方根误差小于0.1°,可以服务于视觉导航算法和综合任务方案在无人平台上的性能评估.

Abstract In the traditional public area pedestrian flow monitoring, a lot of manpower and material resources are needed. Only relying on people’s eyes to identify and count the number of people in the video under monitoring is not only inefficient, but also has errors in accuracy and wastes a certain amount of energy and financial resources. In the rapid development trend of artificial intelligence, this system starts from the perspective of real-time monitoring, with the maturity and development of machine vision technology, and creates a cloud monitoring platform based on JavaWeb technology. The pedestrian detection technology is applied to the pedestrian flow monitoring, which provides a feasible scheme for accurate and fast pedestrian flow monitoring and warning, and develops a system to help public area managers and government departments to deal with the crowd gathering situation, analyze and monitor the crowd density, and give a quick warning to the crowd outbreak and gathering place.

野外基线是测绘科学技术领域中特殊的长度实物标准,可用来检定全站仪等光电测距仪的加、乘常数,为确保检定结果的真实、准确、可靠,定期进行野外基线的溯源是必不可少的.目前我国主要采用24 m因瓦尺或高精度光电测距仪2种方法实现野外基线的溯源,已有大量实验事实表明,我国野外基线的量值传递与国外仍存在着不一致的问题.随着我国制造业的快速发展和“中国制造2025”的提出,传统的野外基线溯源技术很难满足越来越高的精度要求,迫切地需要实现野外基线精密测距.结合国内外研究现状,对光干涉法、24 m因瓦尺法和高精度光电测距法进行总结,深入分析了3种野外基线溯源方式的优缺点.最后,对我国未来的基线场建设提出了一些思考与建议.

目前,使用精密测距仪μ-base测量野外基线时,国内测绘单位测量气象元素往往采用求测站和镜站平均值的方法.为提高测距准确度,结合推导出的μ-base测距仪气象改正公式并依据误差传播定律,对气象代表性误差进行分析,设计了环境参数自动采集系统,在基线沿线布设温湿度、气压传感器,精确采集气象参数.对于普遍采用分段线性插值计算测线温度的方法,提出一种基于径向基神经网络的温度曲线逼近方法.结果表明,曲线整体的平滑性和曲率渐变性较好,更加符合地表温度分布规律,具有良好的效果.

针对现有三轴磁强计误差校正方法存在计算量大、依赖外界参考信息、不能在线校正等问题,提出一种基于递推最小二乘的误差在线自校正方法.根据Poisson方程对磁场测量模型的描述,导出磁场矢量误差校正模型;基于椭球假设理论,建立校正模型与椭球曲面方程系数之间的关系;推导了基于递推最小二乘的椭球方程系数在线辨识的实现过程,进而求得误差校正参数.实验结果表明:提出的方法能有效校正软磁和硬磁效应引起的数据畸变;采样点磁场强度最大波动幅度由67.1128μT降低至14.0648μT,误差标准差由15.8287μT降低至6.3451μT,适用于无外部参考基准下三轴磁强计的误差自动校正.

Measurement and control software of super-station is the core part of super-station and it is also the important part of realizing super-station function. Aiming at the functional characteristics of the super-station, this paper designs the measurement and control software of the super-station and implements it by programming, focusing on the three aspects of the design requirements of the super-station. The interface connection between the measurement and control software and each unit of the super-station is solved in the design. In the aspect of the realization of measurement and control function, the design scheme and flow chart are defined, and the control instruction analysis and interface design are carried out. In the database module, by transplanting FATFS file system, the operation of adding, deleting, modifying and checking measurement data can be realized in the form of files. Finally, we summarize the main work and shortcomings of the design.

This paper introduces in detail methods about calibration length baseline based on mu-base distance meter, and errors in the calculation of oblique distance model; the measurement of meteorological parameters and the correction of height difference are analyzed quantitatively by means of equation derivation. Through the calibration results of mu-base and invar tape, systematic differences between the two measurement methods were found out at the Yellow River Embankment Length Baseline in Zhengzhou. Based on a TS30 total station instrument with stable performance, according to the calculation results of "analytical method" and "comparative method", the accuracy of the two calibration results between mu-base and invar tape is compared and evaluated. In the end, some suggestions to improve the accuracy of using mu-base to calibrate length baseline are given.

UWB (Ultra-wideband) technology is a very promising indoor positioning method because of the advantages of anti-multipath and high precision. For UWB positioning, at least three distance measurements between the base station and a mobile station are required. In practical applications, the signal loss of base stations often occurs due to occlusion and other factors, then resulting in the failure of positioning. To solve this problem, a tightly-coupled method which combines UWB range and measurements of MIMU (Micro Inertial Measurement Unit) is proposed in this paper. The integration positioning system can work continuously during the loss of UWB signals with advantages of INS (Inertial Navigation System) technology’s autonomy and high precision in a short time. A tightly-coupled filter based on EKF (Extended Kalman Filter) is designed, and the integration system is implemented. The simulation based on Matlab and experiments in the indoor testing scene based on the Optitrack system are carried out respectively. The results show that the UWB/MIMU tightly-coupled system can solve the problem of positioning discontinuity caused by signal loss of UWB base stations, and the precision of dynamic positioning is better than 0.3 meters in complex indoor environments.

为解决基于视觉方法测量柔性结构模态测量范围小、无法同时测量大型柔性结构纵深方向多个节点的振动情况的问题,提出基于单目红外的远距离多测点振动测量方法.在单红外相机的基础上,加装带通滤光片、红外LED等设备构成测量系统.利用离焦成像的原理对不同距离段内的LED灯组同时成像,对测量结果使用卡尔曼滤波平滑减小误差,滤波后定位精度为0.5 mm,重复性定位精度优于0.2 mm.通过在不同距离布设不锈钢尺模拟大型柔性结构,测量及实验计算结果与固有频率理论值进行比较相对误差最小为2.22％,计算过程中,滤波后数据有效地排除了干扰峰值.实验结果表明,所提测量方法简单有效,可应用于大型柔性结构模态检测等领域.

针对单目视觉存在尺度不确定性,无法准确估计移动机器人位姿的问题,提出一种具备尺度恢复的单目视觉里程计方法.通过建立局部地图解决基于参考帧/当前帧的位姿估计方式过分依赖参考帧的问题;为了提高位姿估计精度,在使用滑动窗口控制计算量的基础上建立位姿与地图3D点的图优化模型进行光束法平差;最后基于平面假设成立和相机高度已知的条件下,通过图像RoI区域的稠密匹配,建立非线性优化模型解得尺度因子,从而实现绝对尺度恢复.实验结果表明,该方法能稳定地进行位姿估计并且有效地解决了单目视觉存在的尺度不确定性问题,可用于真实三维场景中移动机器人的位姿估计.

圆心检测中,圆的边缘缺损会影响圆心检测的精度.霍夫圆圆心检测具有良好的稳定性,但是精度稍低;最小二乘圆拟合,具有较高的精度,但是易受噪声点的干扰.针对这种情况,本文提出了一种基于霍夫圆的亚像素缺损圆圆心检测方法.先利用霍夫圆检测出粗略圆心坐标及半径,再利用轮廓检测技术提取出所有轮廓点的坐标,并根据霍夫圆检测出圆心及半径对轮廓坐标进行筛选,保留在圆上的轮廓坐标,并利用这些轮廓点坐标进行最小二乘圆拟合求出圆心.实验结果表明,该方法适用性较广,简单可靠,精度高.

在对圆形目标进行图像处理以实现精密测量的各种任务中,圆心定位的精度直接决定了测量结果的精度.目前,成熟的圆心定位算法包括重心法、Hough变换法、高斯拟合法、圆拟合法以及椭圆拟合法等.这些常用算法各具优缺点,因此有不同的适用场景.本文提出了一种新的研究思路,利用高斯过程模型解决圆心定位的问题,并且提出了分别基于标准圆和椭圆方程的高斯过程圆心定位算法.实验结果表明,高斯过程标准圆中心定位算法与圆拟合算法的精度相当,而且在残缺圆的圆心定位中抗噪能力更好,精度更高;高斯过程椭圆中心定位算法比椭圆拟合法的抗噪能力更强,精度更高.

针对目前视觉同步定位与建图(V-SLAM)动态场景适应性差的问题,提出了一种结合查找表和金字塔LK光流法的改进单目视觉同步定位与建图方法.该方法通过相邻帧的视差估计图像的移动方向,利用金字塔LK光流法获取光流信息,根据光流大小与阈值的比较检测移动物体,最终结合查找表减少移动物体对后续定位与建图的影响.实验证明该方法可以有效提高定位精度,减少图像的冗余信息和动态目标的影响.能满足视觉SLAM在室内场景下的实时需求,具有较高的准确性、鲁棒性和适应性.