Uncertainty comparison of three visual odometry systems in different operative conditions

An experimental comparison among visual odometry systems using lenses with three different focal lengths (an ultra wide angle, a medium wide angle and a telephoto lens) is presented. For each focal length, several translational and rotational tests are performed, taking into account and analyzing different positions of the system inside the laboratory. The influence of several operative parameters is analyzed, highlighting their effect on the visual odometry systems equipped with different lenses. Experimental errors and uncertainties obtained by the three systems are compared.     

Contour-based 3D motion recovery while zooming

This paper considers the problem of 3D motion recovery from a sequence of monocular images while zooming. Unlike the common trend based on point matches, the proposed method relies on the deformation of an active contour fitted to a reference object. We derive the relation between the contour deformation and the 3D motion components, assuming time-varying focal length and principal point. This relation allows us to present a method to extract the rotation matrix and the scaled translation along the optical axis.     

基于局部熵的SLAM视觉里程计优化算法

针对移动机器人视觉同步定位与地图创建中由于相机大角度转动造成的帧间匹配失败以及跟踪丢失等问题, 提出了一种基于局部图像熵的细节增强视觉里程计优化算法. 建立图像金字塔, 划分图像块进行均匀化特征提取, 根据图像块的信息熵判断其信息量大小, 将对比度低以及梯度变化小的图像块进行删除, 减小图像特征点计算量. 对保留的图像块进行亮度自适应调整, 增强局部图像细节, 尽可能多地提取能够表征图像信息的局部特征点作为相邻帧匹配以及关键帧匹配的关联依据. 结合姿态图优化方法对位姿累计误差进行局部和全局优化, 进一步提高移动机器人系统性能. 采用TUM数据集测试验证, 由于提取了更能反映物体纹理以及形状的特征属性, 本文算法的运动跟踪成功率最高可提升至60 % 以上, 并且测量的轨迹误差、平移误差以及转动误差都有所降低. 与目前ORB-SLAM2系统相比, 本文提出的算法不但提高了移动机器人视觉定位精度, 而且满足实时SLAM的应用需要.     

Online Detection of State Estimator Performance Degradation via Efficient Numerical Observability Analysis

An efficient observability analysis method is proposed to enable online detection of performance degradation of an optimization-based sliding window visual-inertial state estimation framework. The proposed methodology leverages numerical techniques in nonlinear observability analysis to enable online evaluation of the system observability and indication of the state estimation performance. Specifically, an empirical observability Gramian based approach is introduced to efficiently measure the observability condition of the windowed nonlinear system, and a scalar index is proposed to quantify the average system observability. The proposed approach is specialized to a challenging optimization-based sliding window monocular visual-inertial state estimation formulation and evaluated through simulation and experiments to assess the efficacy of the methodology. The analysis result shows that the proposed approach can correctly indicate degradation of the state estimation accuracy with real-time performance.     

视觉—惯性导航定位技术研究进展

视觉—惯性导航定位技术是一种利用视觉传感器和惯性传感器实现载体的自定位和周围环境感知的无源导航定位方式,可以在全球定位系统（global positioning system,GPS）拒止环境下实现载体6自由度位姿估计。视觉和低精度惯性传感器具有体积小和价格低的优势,得益于二者在导航定位任务中的互补特性,视觉—惯性导航系统（visual inertial navigation system,VINS）引起了极大关注,在移动端的虚拟现实（virtual reality,VR）、增强现实（augmented reality,AR）以及无人系统的自主导航任务中发挥了重要作用,具有重要的理论研究价值和实际应用需求。本文介绍视觉—惯性导航系统,总结概括该系统中初始化、视觉前端处理、状态估计、地图的构建与维护以及信息融合等关键技术的研究进展。对非理想环境下及基于学习方法的视觉—惯性导航定位算法等热点问题进行综述,总结用于算法评测的方法及标准数据集,阐述该技术在实际应用中所面临的主要问题,并针对这些问题对该领域未来的发展趋势进行展望。     

Computer Vision on Mars

Increasing the level of spacecraft autonomy is essential for broadening the reach of solar system exploration. Computer vision has and will continue to play an important role in increasing autonomy of both spacecraft and Earth-based robotic vehicles. This article addresses progress on computer vision for planetary rovers and landers and has four main parts. First, we review major milestones in the development of computer vision for robotic vehicles over the last four decades. Since research on applications for Earth and space has often been closely intertwined, the review includes elements of both. Second, we summarize the design and performance of computer vision algorithms used on Mars in the NASA/JPL Mars Exploration Rover (MER) mission, which was a major step forward in the use of computer vision in space. These algorithms did stereo vision and visual odometry for rover navigation and feature tracking for horizontal velocity estimation for the landers. Third, we summarize ongoing research to improve vision systems for planetary rovers, which includes various aspects of noise reduction, FPGA implementation, and vision-based slip perception. Finally, we briefly survey other opportunities for computer vision to impact rovers, landers, and orbiters in future solar system exploration missions.     

基于Harris角点检测的立体视觉里程计

传统的视觉里程计算法对弱纹理区域和远距离区域的计算精度较低,且算法实时性较差。本文介绍了一种视觉里程计算的可行方法：通过特征提取、特征匹配和运动估计得出车体运动的距离和方向信息,并通过极线几何约束、多帧跟踪和变焦距等方法有效地提高了精度。采用Harris特征点提取和线性与非线性最小二乘结合的运动估计方法,在确保精度的同时提高了算法的实时性。实验表明效果良好。     

基于Accodometry法的两轮自平衡机器人位置估计研究

针对两轮自平衡机器人运行过程中遇到打滑、越障、碰撞等异常事件,测程法进行位置估计失效的情况,提出一种Accodometry方法,通过融合码盘与加速度计数据对位置进行估计,解决了非系统测程法误差对机器人位置估计的影响,降低了加速度计固有漂移的不利影响,提高了两轮自平衡机器人的定位精度.实验验证了Accodometry方法的有效性,结果显示位置误差降为原来的1/4.     

基于在线光度标定的半直接视觉SLAM算法

为解决视觉SLAM（同时定位与地图创建）算法依赖图像亮度而对光照变化场景敏感的问题,提出一种基于在线光度标定的半直接视觉SLAM算法。首先,根据相机成像原理,提出基于光度标定的帧间位姿估计方法,在求解位姿的同时对原始的输入图像进行光度校正。其次,在特征追踪环节采取最近共视关键帧匹配策略,以提升特征点匹配效率。最后,对后端重投影迭代优化策略进行改进,降低光照变化对视觉SLAM算法的精度和鲁棒性的影响。在TUM、EuRoC数据集上的实验结果表明,本算法的轨迹估计精度优于LSD-SLAM和SVO 2.0算法,尤其是在中等难度、高难度的数据集序列上。在真实环境测试中,通过对比本算法与激光方法的轨迹估计结果,证明本算法有效提高了传统视觉SLAM方法在光照不均匀场景下的定位精度与鲁棒性。     

基于视觉惯性融合的半直接单目视觉里程计

针对半直接单目视觉里程计缺乏尺度信息并且在快速运动中鲁棒性较差的缺点,设计了一种融合惯性测量信息的半直接单目视觉里程计,通过IMU(惯性测量单元)信息弥补视觉里程计的缺陷,有效提高跟踪精度与系统鲁棒性.本文联合惯性测量信息与视觉信息进行初始化,较准确地恢复了环境尺度信息.为提高运动跟踪的鲁棒性,提出一种IMU加权的运动先验模型.通过预积分获取IMU的状态估计,根据IMU先验误差调整权重系数,使用IMU先验信息的加权值为前端提供精确的初值.后端构建了紧耦合的图优化模型,融合惯性、视觉以及3维地图点信息进行联合优化,同时在滑动窗口中使用强共视关系作为约束,在消除局部累积误差的同时提高优化效率与优化精度.实验结果表明,本文的先验模型优于匀速运动模型与IMU先验模型,单帧先验误差小于1 cm.后端优化方法改进后,计算效率提高为原来的1.52倍,同时轨迹精度与优化稳定性也得到提高.在EuRoC数据集上进行测试,定位效果优于OKVIS算法,轨迹均方根误差减小为原视觉里程计的1/3.