Search results

The purpose of this paper is to propose a binocular visual odometry algorithm based on the Random Sample Consensus (RANSAC) in visual navigation systems.

The authors propose a novel binocular visual odometry algorithm based on features from accelerated segment test (FAST) extractor and an improved matching method based on the RANSAC. Firstly, features are detected by utilizing the FAST extractor. Secondly, the detected features are roughly matched by utilizing the distance ration of the nearest neighbor and the second nearest neighbor. Finally, wrong matched feature pairs are removed by using the RANSAC method to reduce the interference of error matchings.

The performance of this new algorithm has been examined by an actual experiment data. The results shown that not only the robustness of feature detection and matching can be enhanced but also the positioning error can be significantly reduced by utilizing this novel binocular visual odometry algorithm. The feasibility and effectiveness of the proposed matching method and the improved binocular visual odometry algorithm were also verified in this paper.

This paper presents an improved binocular visual odometry algorithm which has been tested by real data. This algorithm can be used for outdoor vehicle navigation.

A binocular visual odometer algorithm based on FAST extractor and RANSAC methods is proposed to improve the positioning accuracy and robustness. Experiment results have verified the effectiveness of the present visual odometer algorithm.

Terrestrial laser scanning (TLS) point clouds have been widely used in deformation measurement for structures. However, reliability and accuracy of resulting deformation estimation strongly depends on quality of each step of a workflow, which are not fully addressed. This study aims to give insight error of these steps, and results of the study would be guidelines for a practical community to either develop a new workflow or refine an existing one of deformation estimation based on TLS point clouds. Thus, the main contributions of the paper are investigating point cloud registration error affecting resulting deformation estimation, identifying an appropriate segmentation method used to extract data points of a deformed surface, investigating a methodology to determine an un-deformed or a reference surface for estimating deformation, and proposing a methodology to minimize the impact of outlier, noisy data and/or mixed pixels on deformation estimation.

In practice, the quality of data point clouds and of surface extraction strongly impacts on resulting deformation estimation based on laser scanning point clouds, which can cause an incorrect decision on the state of the structure if uncertainty is available. In an effort to have more comprehensive insight into those impacts, this study addresses four issues: data errors due to data registration from multiple scanning stations (Issue 1), methods used to extract point clouds of structure surfaces (Issue 2), selection of the reference surface S_ref to measure deformation (Issue 3), and available outlier and/or mixed pixels (Issue 4). This investigation demonstrates through estimating deformation of the bridge abutment, building and an oil storage tank.

The study shows that both random sample consensus (RANSAC) and region growing–based methods [a cell-based/voxel-based region growing (CRG/VRG)] can be extracted data points of surfaces, but RANSAC is only applicable for a primary primitive surface (e.g. a plane in this study) subjected to a small deformation (case study 2 and 3) and cannot eliminate mixed pixels. On another hand, CRG and VRG impose a suitable method applied for deformed, free-form surfaces. In addition, in practice, a reference surface of a structure is mostly not available. The use of a fitting plane based on a point cloud of a current surface would cause unrealistic and inaccurate deformation because outlier data points and data points of damaged areas affect an accuracy of the fitting plane. This study would recommend the use of a reference surface determined based on a design concept/specification. A smoothing method with a spatial interval can be effectively minimize, negative impact of outlier, noisy data and/or mixed pixels on deformation estimation.

Due to difficulty in logistics, an independent measurement cannot be established to assess the deformation accuracy based on TLS data point cloud in the case studies of this research. However, common laser scanners using the time-of-flight or phase-shift principle provide point clouds with accuracy in the order of 1–6 mm, while the point clouds of triangulation scanners have sub-millimetre accuracy.

This study aims to give insight error of these steps, and the results of the study would be guidelines for a practical community to either develop a new workflow or refine an existing one of deformation estimation based on TLS point clouds.

The results of this study would provide guidelines for a practical community to either develop a new workflow or refine an existing one of deformation estimation based on TLS point clouds. A low-cost method can be applied for deformation analysis of the structure.

Although a large amount of the studies used laser scanning to measure structure deformation in the last two decades, the methods mainly applied were to measure change between two states (or epochs) of the structure surface and focused on quantifying deformation-based TLS point clouds. Those studies proved that a laser scanner could be an alternative unit to acquire spatial information for deformation monitoring. However, there are still challenges in establishing an appropriate procedure to collect a high quality of point clouds and develop methods to interpret the point clouds to obtain reliable and accurate deformation, when uncertainty, including data quality and reference information, is available. Therefore, this study demonstrates the impact of data quality in a term of point cloud registration error, selected methods for extracting point clouds of surfaces, identifying reference information, and available outlier, noisy data and/or mixed pixels on deformation estimation.

With the improvement of modern aircraft requirements for safety, long life and economy, higher quality aircraft assembly is needed. However, due to the manufacturing and assembly errors of the posture adjustment mechanism (PAM) used in the digital assembly of aircraft large component (ALC), the posture alignment accuracy of ALC is difficult to be guaranteed, and the posture adjustment stress is easy to be generated. Aiming at these problems, this paper aims to propose a calibration method of redundant actuated parallel mechanism (RAPM) for posture adjustment.

First, the kinematics model of the PAM is established, and the influence of the coupling relationship between the axes of the numerical control locators (NCL) is analyzed. Second, the calibration method based on force closed-loop feedback is used to calibrate each branch chain (BC) of the PAM, and the solution of kinematic parameters is optimized by Random Sample Consensus (RANSAC). Third, the uncertainty of kinematic calibration is analyzed by Monte Carlo method. Finally, a simulated posture adjustment system was built to calibrate the kinematics parameters of PAM, and the posture adjustment experiment was carried out according to the calibration results.

The experiment results show that the proposed calibration method can significantly improve the posture adjustment accuracy and greatly reduce the posture adjustment stress.

In this paper, a calibration method based on force feedback is proposed to avoid the deformation of NCL and bracket caused by redundant driving during the calibration process, and RANSAC method is used to reduce the influence of large random error on the calibration accuracy.

The purpose of this study is to propose a novel plane-based mapping method for legged-robot navigation in a stairway environment.

The approach implemented in this study estimates a plane for each step of a stairway using a weighted average of sensor measurements and predictions. It segments planes from point cloud data via random sample consensus (RANSAC). The prediction uses the regular structure of a stairway. When estimating a plane, the algorithm considers the errors introduced by the distance sensor and RANSAC, in addition to stairstep irregularities, by using covariance matrices. The plane coefficients are managed separately with the data structure suggested in this study. In addition, this data structure allows the algorithm to store the information of each stairstep as a single entity.

In the case of a stairway environment, the accuracy delivered by the proposed algorithm was higher than those delivered by traditional mapping methods. The hardware experiment verified the accuracy and applicability of the algorithm.

The proposed algorithm provides accurate stairway-environment mapping and detailed specifications of each stairstep. Using this information, a legged robot can navigate and plan its motion in a stairway environment more efficiently.

The purpose of this paper is to present a novel registration method for augmented reality (AR) systems based on robust estimation of trifocal tensor using point and line correspondence simultaneously.

The proposed method distinguishes itself in following three ways: first, to establish the world coordinate system, the restriction that the four specified points must form an approximate square is relaxed, the only requirement is that these four points should not be collinear. Second, besides feature points, line segments are also used to calculate the needed trifocal tensors. The registration process can still be achieved even without the use of feature points. Third, to estimate trifocal tensors precisely, progressive sample consensus (PROSAC) is used instead of random sample consensus to remove outliers.

As shown in the experiments, the proposed method really enhances the usability of this system. To calculate trifocal tensor, a PROSAC based algebraic minimization algorithm is put forward which improves the accuracy and reduces the computation complexity.

In current system, it is stipulated that there is no large rotation of the user's head relative to the registration scenes, because the NCC will degrade when there is a large rotation between images.

A more robust feature matching strategy is needed. Treating feature matching as a classification problem may be a good choice.

This paper presents a novel registration approach for AR system.

Positioning is a key task in most field robotics applications but can be very challenging in GPS‐denied or high‐slip environments. The purpose of this paper is to describe a visual odometry strategy using only one camera in country roads.

This monocular odometery system uses as input only those images provided by a single camera mounted on the roof of the vehicle and the framework is composed of three main parts: image motion estimation, ego‐motion computation and visual odometry. The image motion is estimated based on a hyper‐complex wavelet phase‐derived optical flow field. The ego‐motion of the vehicle is computed by a blocked RANdom SAmple Consensus algorithm and a maximum likelihood estimator based on a 4‐degrees of freedom motion model. These as instantaneous ego‐motion measurements are used to update the vehicle trajectory according to a dead‐reckoning model and unscented Kalman filter.

The authors' proposed framework and algorithms are validated on videos from a real automotive platform. Furthermore, the recovered trajectory is superimposed onto a digital map, and the localization results from this method are compared to the ground truth measured with a GPS/INS joint system. These experimental results indicate that the framework and the algorithms are effective.

The effective framework and algorithms for visual odometry using only one camera in country roads are introduced in this paper.

In corridor environments, human-following robot encounter difficulties when the target turning around at the corridor intersections, as walls may cause complete occlusion. This paper aims to propose a collision-free following system for robot to track humans in corridors without a prior map.

In addition to following a target and avoiding collisions robustly, the proposed system calculates the positions of walls in the environment in real-time. This allows the system to maintain a stable tracking of the target even if it is obscured after turning. The proposed solution is integrated into a four-wheeled differential drive mobile robot to follow a target in a corridor environment in real-world.

The experimental results demonstrate that the robot equipped with the proposed system is capable of avoiding obstacles and following a human target robustly in the corridors. Moreover, the robot achieves a 90% success rate in maintaining a stable tracking of the target after the target turns around a corner with high speed.

This paper proposes a human target following system incorporating three novel features: a path planning method based on wall positions is introduced to ensure stable tracking of the target even when it is obscured due to target turns; improvements are made to the random sample consensus (RANSAC) algorithm, enhancing its accuracy in calculating wall positions. The system is integrated into a four-wheeled differential drive mobile robot effectively demonstrates its remarkable robustness and real-time performance.

This paper aims to quickly obtain an accurate and complete dense three-dimensional map of indoor environment with lower cost, which can be directly used in navigation.

This paper proposes an improved ORB-SLAM2 dense map optimization algorithm. This algorithm consists of three parts: ORB feature extraction based on improved FAST-12, feature point extraction with progressive sample consensus (PROSAC) and the dense ORB-SLAM2 algorithm for mapping. Here, the dense ORB-SLAM2 algorithm adds LoopClose optimization thread and dense point cloud map and octree map construction thread. The dense map is computationally expensive and occupies a large amount of memory. Therefore, the proposed method takes higher efficiency, voxel filtering can reduce the memory while ensuring the density of the map and then use the octree format to store the map to further reduce memory.

The improved ORB-SLAM2 algorithm is compared with the original ORB-SLAM2 algorithm, and the experimental results show that the map through improved ORB-SLAM2 can be directly used in navigation process with higher accuracy, shorter tracking time and smaller memory.

The improved ORB-SLAM2 algorithm can obtain a dense environment map, which ensures the integrity of data. The comparisons of FAST-12 and improved FAST-12, RANSAC and PROSAC prove that the improved FAST-12 and PROSAC both make the feature point extraction process faster and more accurate. Voxel filter helps to take small storage memory and low computation cost, and octree map construction on the dense map can be directly used in navigation.

Simultaneous localization and map building (SLAM), as a state estimation problem, is a prerequisite for solving the problem of autonomous vehicle motion in unknown environments. Existing algorithms are based on laser or visual odometry; however, the lidar sensing range is small, the amount of data features is small, the camera is vulnerable to external conditions and the localization and map building cannot be performed stably and accurately using a single sensor. This paper aims to propose a laser three dimensions tightly coupled map building method that incorporates visual information, and uses laser point cloud information and image information to complement each other to improve the overall performance of the algorithm.

The visual feature points are first matched at the front end of the method, and the mismatched point pairs are removed using the bidirectional random sample consensus (RANSAC) algorithm. The laser point cloud is then used to obtain its depth information, while the two types of feature points are fed into the pose estimation module for a tightly coupled local bundle adjustment solution using a heuristic simulated annealing algorithm. Finally, the visual bag-of-words model is fused in the laser point cloud information to establish a threshold to construct a loopback framework to further reduce the cumulative drift error of the system over time.

Experiments on publicly available data sets show that the proposed method in this paper can match its real trajectory well. For various scenes, the map can be constructed by using the complementary laser and vision sensors, with high accuracy and robustness. At the same time, the method is verified in a real environment using an autonomous walking acquisition platform, and the system loaded with the method can run well for a long time and take into account the environmental adaptability of multiple scenes.

A multi-sensor data tight coupling method is proposed to fuse laser and vision information for optimal solution of the positional attitude. A bidirectional RANSAC algorithm is used for the removal of visual mismatched point pairs. Further, oriented fast and rotated brief feature points are used to build a bag-of-words model and construct a real-time loopback framework to reduce error accumulation. According to the experimental validation results, the accuracy and robustness of the single-sensor SLAM algorithm can be improved.

The purpose of this paper is to describe the procedure for near-automation of the most commonly used manual georeferencing technique in a desktop GIS environment for historic aerial photographs strip in library archives.

Most of the archived historic aerial photography consists of series of aerial photographs that overlap to some extent, as the optimal overlap ratio is known as 60 percent by photogrammetric standard. Therefore, conjugate points can be detected for the overlapping area. The first image was georeferenced manually by six-parameter affine transformation using 2013 National Agriculture Imagery Program images as ground truths. Then, conjugate points were detected in the first and second images using Speeded Up Robust Features and Random Sample Consensus. The ground space coordinates of conjugate points were estimated using the first image’s six parameters. Then the second image’s six parameters were calculated using conjugate points’ ground space coordinates and pixel coordinates in the second image. This procedure was repeated until the last image was georeferenced. However, error accumulated as the number of photographs increased. Therefore, another six-parameter affine transformation was implemented using control points in the first, middle, and last images. Finally, the images were warped using open source GIS tools.

The result shows that historic aerial strip collections can be georeferenced with far less time and labor using the technique proposed compared with the traditional manual georeferencing technique in a desktop GIS environment.

The suggested approach will promote the usage of historic aerial photographs for various scientific purposes including land use and land cover change detection, soil erosion pattern recognition, agricultural practices change analysis, environmental improvement assessment, and natural habitat change detection.

Most commonly used georeferencing procedures for historic aerial photographs in academic libraries require significant time and effort for manual measurement of conjugate points in the object images and the ground truth images. By maximizing the automation of georeferencing procedures, the suggested approach will save significant time and effort of library workforce.

With the suggested approach, large numbers of historic aerial photographs can be rapidly georeferenced. This will allow libraries to provide more geospatial data to scientific communities.

This is a unique approach to rapid georeferencing of historic aerial photograph strips.