Visual odometry (VO) is a key technology for estimating camera motion from captured images. In this paper, we propose a novel RGB-D visual odometry by constructing and matching features at the superpixel level that represents better adaptability in different environments than state-of-the-art solutions. Superpixels are content-sensitive and perform well in information aggregation. They could thus characterize the complexity of the environment. Firstly, we designed the superpixel-based feature SegPatch and its corresponding 3D representation MapPatch. By using the neighboring information, SegPatch robustly represents its distinctiveness in various environments with different texture densities. Due to the inclusion of depth measurement, the MapPatch constructs the scene structurally. Then, the distance between SegPatches is defined to characterize the regional similarity. We use the graph search method in scale space for searching and matching. As a result, the accuracy and efficiency of matching process are improved. Additionally, we minimize the reprojection error between the matched SegPatches and estimate camera poses through all these correspondences. Our proposed VO is evaluated on the TUM dataset both quantitatively and qualitatively, showing good balance to adapt to the environment under different realistic conditions.

Although conflict archaeology is now well established, the archaeological remains of many specific military confrontations are still to be explored. This article reports the results of fieldwork to document the site of the Battle of the Bulge (16 December 1944–25 January 1945). The authors use drone-mounted 1m-resolution LiDAR and very high-resolution simultaneous localisation and mapping (SLAM) methods to reveal more than 940 features within the forested Ardennes landscape, many of which were subsequently visited and confirmed. As well as highlighting the potential of the LiDAR-SLAM method, deployed here (both in this geographic region and in conflict archaeology) for the first time, the survey results emphasise the need for a debate on managing the heritage of a key modern conflict landscape in Europe.

In this paper, we propose a comparison of open-source LiDAR and Inertial Measurement Unit (IMU)-based Simultaneous Localization and Mapping (SLAM) approaches for 3D robotic mapping. The analyzed algorithms are often exploited in mobile robotics for autonomous navigation but have not been evaluated in terms of 3D reconstruction yet. Experimental tests are carried out using two different autonomous mobile platforms in three test cases, comprising both indoor and outdoor scenarios. The 3D models obtained with the different SLAM algorithms are then compared in terms of density, accuracy, and noise of the point clouds to analyze the performance of the evaluated approaches. The experimental results indicate the SLAM methods that are more suitable for 3D mapping in terms of the quality of the reconstruction and highlight the feasibility of mobile robotics in the field of autonomous mapping.

This paper presents a new feature based approach for place recognition using 2D LiDAR (Light Detection And Ranging) data. The main contribution lies in the mapping process. It includes a keypoint selection strategy to model places with persistent keypoints from concatenated LiDAR scans. Our objective is to achieve map data reduction while maintaining good place recognition performace. LiDAR scans are concatenated with a registration algorithm and keypoints are extracted from each scan. Based on a regular grid, our approach measures the occurrence of similar keypoints in each region of interest defined by a grid cell. Only keypoints with occurrences beyond a threshold, qualified as stable keypoints, are kept in the place model called submap. The environment is therefore modeled as a collection of submaps, which constitutes the global map. Place recognition consists in submap matching followed by a two steps geometric verification. In the first stage, optimal parameters are set using corrected data. Mapping parameters satisfy six criteria among which is the spatial distribution distance, which represents another contribution of our work. It gives a measure on how well keypoints are distributed in space. Place recognition optimal parameters are set through global localization. In the second stage, the approach is evaluated using raw data in the contexts of global localization, and loop closure detection in a SLAM framework. The results obtained using popular real data sets show that our approach achieves significant map data reduction (up to $92\%$ ) while maintaining good place recognition performance, comparable to state-of-the-art methods.

Localization based on visual natural landmarks is one of the state-of-the-art localization methods for automated vehicles that is, however, limited in fast motion and low-texture environments, which can lead to failure. This paper proposes an approach to solve these limitations with an extended Kalman filter (EKF) based on a state estimation algorithm that fuses information from a low-cost MEMS Inertial Measurement Unit and a Time-of-Flight camera. We demonstrate our results in an indoor environment. We show that the proposed approach does not require any global reflective landmark for localization and is fast, accurate, and easy to use with mobile robots.

An improved FastSLAM based on the robust square-root cubature Kalman filter (RSRCKF) with partial genetic resampling is proposed in this paper. In the proposed method, RSRCKF is used to design the proposal distribution of FastSLAM and to estimate environment landmarks. The proposed method does not require a priori knowledge of the noise statistics. In addition, to increase diversity, it uses the genetic operators-based strategy to further improve the particle diversity. In fact, a partial genetic resampling operation is carried out to maintain the diversity of particles. The proposed method is compared with other methods via simulation and experimental data. It can be seen from the results that the proposed method provides significantly more accurate and robust estimation results compared with other methods even with fewer particles and unknown a priori. In addition, the consistency of the proposed method is better than that of other methods.

This article proposes a method for incremental data dimensionality reduction in loop closure detection for robotic autonomous navigation. The approach uses dominant eigenvector concept for: (a) spectral description of visual datasets and (b) representation in low dimension. Unlike most other papers on data dimensionality reduction (which is done in batch mode), our method combines a sliding window technique and coordinate transformation to achieve dimensionality reduction in incremental data. Experiments in both simulated and real scenarios were performed and the results are suitable.

Conventional simultaneous localization and mapping (SLAM) has concentrated on two-dimensional (2D) map building. To adapt it to urgent search and rescue (SAR) environments, it is necessary to combine the fast and simple global 2D SLAM and three-dimensional (3D) objects of interest (OOIs) local sub-maps. The main novelty of the present work is a method for 3D OOI reconstruction based on a 2D map, thereby retaining the fast performances of the latter. A theory is established that is adapted to a SAR environment, including the object identification, exploration area coverage (AC), and loop closure detection of revisited spots. Proposed for the first is image optical flow calculation with a 2D/3D fusion method and RGB-D (red, green, blue + depth) transformation based on Joblove–Greenberg mathematics and OpenCV processing. The mathematical theories of optical flow calculation and wavelet transformation are used for the first time to solve the robotic SAR SLAM problem. The present contributions indicate two aspects: (i) mobile robots depend on planar distance estimation to build 2D maps quickly and to provide SAR exploration AC; (ii) 3D OOIs are reconstructed using the proposed innovative methods of RGB-D iterative closest points (RGB-ICPs) and 2D/3D principle of wavelet transformation. Different mobile robots are used to conduct indoor and outdoor SAR SLAM. Both the SLAM and the SAR OOIs detection are implemented by simulations and ground-truth experiments, which provide strong evidence for the proposed 2D/3D reconstruction SAR SLAM approaches adapted to post-disaster environments.

An accurate 3D model of an outdoor scene can be used in many different scenarios of precision agriculture, for instance to analyse the silhouette of a tree crown canopy for precision spraying, to count fruit for fruit yield prediction or to simply navigate a vehicle between the plant rows. Instead of using stereovision, limited by the problems of different light intensities, or by using expensive multi-channel 3D range finder (LIDAR scanner), limited by the number of channels, this work investigates the possibility of using two single channel LIDAR scanners mounted on a small robot to allow a real-time 3D object reconstruction of the robot environment. The approach used readings captured by two LIDAR scanners, SICK LMS111 and SICK TiM310, where the first one was scanning horizontally and the second one vertically. In order to correctly map the 3D points of the readings from the vertical sensor into a 3D space, a custom SLAM algorithm based on image registration techniques was used to calculate the new positions of the robot. The approach was tested in an indoor and outdoor environment, proving its accuracy with an error rate of 0.02 m±0.02 m for vertical and −0.01 m±0.13 m for the horizontal plane.

We present a real time 3D SLAM system for texture-less scenes using only depth information provided by a low cost RGB-D sensor. The proposed method is based on a novel informative sampling scheme that extracts points carrying the most useful 3D information for registration. The aim of the proposed sampling technique is to informatively sample a point cloud into a subset of points based on their 3D information. The flatness of a point is measured by applying a rank order statistics based robust segmentation method to surface normals in its local vicinity. The extracted keypoints from sequential frames are then matched and a rank order statistics based robust estimator is employed to refine the matches and estimate a rigid-body transformation between the frames. Experimental evaluations show that the proposed keypoint extraction method is highly repeatable and outperforms the state of the art methods in terms of accuracy and repeatability. We show that the performance of the registration algorithm is also comparable to other well-known methods in texture-less environments.

In the context of topological mapping, the automatic segmentation of an environment into meaningful and distinct locations is still regarded as an open problem. This paper presents an algorithm to extract places online from image sequences based on the algebraic connectivity of graphs or Fiedler value, which provides an insight into how well connected several consecutive observations are. The main contribution of the proposed method is that it is a theoretically supported alternative to tuning thresholds on similarities, which is a difficult task and environment dependent. It can accommodate any type of feature detector and matching procedure, as it only requires non-negative similarities as input, and is therefore able to deal with descriptors of variable length, to which statistical techniques are difficult to apply. The method has been validated in an office environment using exclusively visual information. Two different types of features, a bag-of-words model built from scale invariant feature transform (SIFT) keypoints, and a more complex fingerprint based on vertical lines, color histograms, and a few Star keypoints, are employed to demonstrate that the method can be applied to both fixed and variable length descriptors with similar results.

This paper presents what is termed as the supervoxel normal distributions transform (SV-NDT), a novel three-dimensional (3-D) registration algorithm which improves the performance of the three-dimensional normal distributions transform (3-D NDT) significantly. The 3-D NDT partitions a model scan using a 3-D regular grid. Generating normal distributions using the 3-D regular grid causes considerable information loss because the 3-D regular grid does not use any information pertaining to the local surface structures of the model scan. The best type of surface (the constituent unit of each scan) for modeling with one normal distribution is known to be the plane. The SV-NDT reduces the loss of information using a supervoxel-generating algorithm at the partitioning stage. In addition, it uses the information of the local surface structures from the data scan by replacing the Euclidean distance with a function that uses local geometries as well as the Euclidean distance when each point in the data scan is matched to the corresponding normal distribution. Experiments demonstrate that the use of the supervoxel-generating algorithm increases the modeling accuracy of the normal distributions and that the proposed 3-D registration algorithm outperforms the 3-D NDT and other widely used 3-D registration algorithms in terms of robustness and speed on both synthetic and real-world datasets. Additionally, the effect of changing the function to create correspondences is also verified.

Localization and mapping in indoor environments, such as airports and hospitals, are key tasks for almost every robotic platform. Some researchers suggest the use of Range-Only (RO) sensors based on WiFi (Wireless Fidelity) technology with SLAM (Simultaneous Localization And Mapping) techniques to solve both problems. The current state of the art in RO SLAM is mainly focused on the filtering approach, while the study of smoothing approaches with RO sensors is quite incomplete. This paper presents a comparison between filtering algorithms, such as EKF and FastSLAM, and a smoothing algorithm, the SAM (Smoothing And Mapping). Experimental results are obtained in indoor environments using WiFi sensors. The results demonstrate the feasibility of the smoothing approach using WiFi sensors in an indoor environment.

This paper presents a concept of a laser scanner framework designed for obstacle avoidance used on mini fixed-wing unmanned aerial vehicles (UAVs) flying in outdoor environments. The innovation is a conical field of view that guarantees tri-dimensional (3D) obstacle detection and localization at any pitch or roll angle. This advantage is very important for the case of fixed-wing UAV flights where the attitude is changing rapidly. Measurement sequences create a map that is represented by a circular grid with the center fitted to the x-axis of the UAV's body, lying in the plane normal to the velocity vector and projected in the front of UAV. This means that the map cells contain differences between the safety zone radius and distances acquired from area in close proximity to the flight path. Actual UAV attitude can be compensated by rotation and shift of two masks of gains that are applied to the map to determine pitch and roll commands. Results of the simulation research conducted on the designed concept are very promising, as they present a combination of lateral and vertical obstacle avoidance. Based on the experience with laser rangefinders operating on a real UAV, it can be convincingly determined that the concept of the laser scanner is able to be brought into reality.

In SLAM (simultaneous localization and mapping), the topological paradigm provides a more natural and compact solution that scales better with the size of the environment. Computer vision has always been regarded as the ideal sensor technology for topological feature extraction and description and several methods have been proposed in the literature, but they are either time-consuming, require plenty of different sensors, or are very sensitive to perceptual aliasing, all of which limit their application scope.

This paper presents a fast-to-compute collection of features extracted from monocular images, and an adaptive matching procedure for location identification in structured indoor environments inspired by the natural language processing field. Although only dominant vertical lines, color histograms, and a reduced number of keypoints are employed in this paper, the matching framework introduced allows for the incorporation of almost any other type of feature. The results of the experiments carried out in a home and an office environment suggest that the proposed method could be used for real-time topological scene recognition even if the environment changes moderately over time. Due to the combination of complementary features, high precision can be achieved within reasonable computation time by using weaker but faster descriptors.

We propose a distributed algorithm for estimating the 3D pose (position and orientation) of multiple robots with respect to a common frame of reference when Global Positioning System is not available. This algorithm does not rely on the use of any maps, or the ability to recognize landmarks in the environment. Instead, we assume that noisy relative measurements between pairs of robots are intermittently available, which can be any one, or combination, of the following: relative pose, relative orientation, relative position, relative bearing, and relative distance. The additional information about each robot's pose provided by these measurements are used to improve over self-localization estimates. The proposed method is similar to a pose-graph optimization algorithm in spirit: pose estimates are obtained by solving an optimization problem in the underlying Riemannian manifold $(SO(3)\times{\mathcal R}^3)^{n(k)}$. The proposed algorithm is directly applicable to 3D pose estimation, can fuse heterogeneous measurement types, and can handle arbitrary time variation in the neighbor relationships among robots. Simulations show that the errors in the pose estimates obtained using this algorithm are significantly lower than what is achieved when robots estimate their pose without cooperation. Results from experiments with a pair of ground robots with vision-based sensors reinforce these findings. Further, simulations comparing the proposed algorithm with two state-of-the-art existing collaborative localization algorithms identify under what circumstances the proposed algorithm performs better than the existing methods. In addition, the question of trade-offs between cost (of obtaining a certain type of relative measurement) and benefit (improvement in localization accuracy) for various types of relative measurements is considered.

This paper presents a new monocular SLAM algorithm that uses straight lines extracted from images to represent the environment. A line is parametrized by two pairs of azimuth and elevation angles together with the two corresponding camera centres as anchors making the feature initialization relatively straightforward. There is no redundancy in the state vector as this is a minimal representation. A bundle adjustment (BA) algorithm that minimizes the reprojection error of the line features is developed for solving the monocular SLAM problem with only line features. A new map joining algorithm which can automatically optimize the relative scales of the local maps is used to combine the local maps generated using BA. Results from both simulations and experimental datasets are used to demonstrate the accuracy and consistency of the proposed BA and map joining algorithms.

One of the main challenges in robotics is navigating autonomously through large, unknown, and unstructured environments. Simultaneous localization and mapping (SLAM) is currently regarded as a viable solution for this problem. As the traditional metric approach to SLAM is experiencing computational difficulties when exploring large areas, increasing attention is being paid to topological SLAM, which is bound to provide sufficiently accurate location estimates, while being significantly less computationally demanding. This paper intends to provide an introductory overview of the most prominent techniques that have been applied to topological SLAM in terms of feature detection, map matching, and map fusion.

In this work, an optimal maneuverability strategy for car-like unmanned vehicles operating in restricted environments is presented. The maneuverability strategy is based on a path planning algorithm that uses the environment information to plan a safe, feasible and optimum path for the unmanned mobile robot. The environment information is obtained by means of a simultaneous localization and mapping (SLAM) algorithm. The SLAM algorithm uses the sensors' information to build a map of the surrounding environment. A Monte Carlo sampling technique is used to find an optimal and safe path within the environment based on the SLAM information. The objective of the planning is to safely reach a desired orientation in a bounded space. Theoretical demonstrations and real-time experimental results (in indoor and outdoor environments) are also presented in this work.