Vision Based Navigation

Urban Air Mobility (UAM) is poised to transform transportation within densely populated cities by leveraging the third dimension-airspace. Vertical TakeOff and Landing (VTOL) and hybrid VTOL-cruise air vehicles are the most compelling and thus popular configurations for UAM applications due to their operational versatility. However, path planning and guidance in such highly dynamic and GNSS-denied environments are likely to produce suboptimal path and erratic maneuvering commands, due to unreliable positioning and attitude determination. To address this challenge, we propose a robust monocular Visual-Inertial Navigation System (VINS), built upon VINS-Mono. Our approach models the uncertainty in both feature tracking and pose estimation, incorporating this into a Mahalanobis-distance-based outlier rejection strategy using a chi-square test. This enables effective suppression of dynamic features, thereby enhancing the robustness of VTOL and hybrid air vehicles navigation and guidance in UAM scenarios. The verification of proposed approach using public datasets demonstrates that consistently high accuracy and robustness can be attained, even under challenging motion profiles.

In urban canyons, the autonomous positioning and obstacle avoidance capabilities of Uncrewed Aerial Vehicles (UAV) are challenged by dense obstructions, causing Non-Line-of-Sight (NLOS) conditions that severely degrade Global Navigation Satellite Systems (GNSS) signals. Current urban UAV operations still rely heavily on manual control and conventional GNSS-Inertial Measurement Unit (IMU) fusion, which lack autonomous obstacle avoidance capabilities and fail to provide precise positioning in GNSS-denied environments. To address these limitations, we propose a novel sensor fusion framework that integrates the Sage-Husa adaptive Kalman Filter (SHKF) with intermittent data from GNSS, IMU, and visual odometry supplied from a Visual-Inertial (VINS)-Fusion algorithm to achieve robust positioning and attitude determination. This enhanced sensor fusion architecture enables a 3DVFH*-based local planner to execute real-time obstacle avoidance and path planning toward the intended destinations. Preliminary verification activities conducted in a high-fidelity Gazebo simulation demonstrate the effective performance of the proposed method in terms of positioning and attitude determination accuracy, while also supporting the optimal reliability and efficiency of the considered obstacle avoidance functionality.

This article considers a low-cost and light weight platform for the task of autonomous flying for inspection in underground mine tunnels. The main contribution of this paper is integrating simple, efficient and well-established methods in the computer vision community in a state of the art vision-based system for Micro Aerial Vehicle (MAV) navigation in dark tunnels. These methods include Otsu's threshold and Moore-Neighborhood object tracing. The vision system can detect the position of low-illuminated tunnels in image frame by exploiting the inherent darkness in the longitudinal direction. In the sequel, it is converted from the pixel coordinates to the heading rate command of the MAV for adjusting the heading towards the center of the tunnel. The efficacy of the proposed framework has been evaluated in multiple experimental field trials in an underground mine in Sweden, thus demonstrating the capability of low-cost and resource-constrained aerial vehicles to fly autonomously through tunnel confined spaces.

A new application of a Kalman-Like robust filtering scheme to Avionics Dead Reckoning is presented. Inertial and air data measurements are used in order to reconstruct the attitude and the body frame velocity of a general aviation aircraft. The proposed estimation algorithm is based on an affine representation of the non linear model of the system, leading to a Kalman filter gain optimizing procedure under a set of Linear Matrix Inequality constraints. The resulting procedure fulfills the standard accuracy requirements for large set of flight condition, and exhibits better robustness characteristics with respect to the model uncertainty than the classic and widely used Extended Kalman Filter.

To be successful in multi-player drone racing, a player must not only follow the race track in an optimal way, but also compete with other drones through strategic blocking, faking, and opportunistic passing while avoiding collisions. Since unveiling one's own strategy to the adversaries is not desirable, this requires each player to independently predict the other players' future actions. Nash equilibria are a powerful tool to model this and similar multi-agent coordination problems in which the absence of communication impedes full coordination between the agents. In this paper, we propose a novel receding horizon planning algorithm that, exploiting sensitivity analysis within an iterated best response computational scheme, can approximate Nash equilibria in real time. We also describe a vision-based pipeline that allows each player to estimate its opponent's relative position. We demonstrate that our solution effectively competes against alternative strategies in a large number of drone racing simulations. Hardware experiments with onboard vision sensing prove the practicality of our strategy. Video of the experiments: . Drone racing has recently become a popular sport with international competitions being held regularly and attracting a growing public . In these races, human pilots directly control the UAVs through a radio transmitter while receiving a first-person-view live stream from an onboard camera. Human racers need years of training to master the advanced navigation and control skills that are required to be successful in this sport. Many of these skills would certainly prove useful for a robot to safely and quickly move through a cluttered environment in, for example, a disaster response scenario. For this reason, drone racing has attracted a significant interest from the scientific community, which led to the first autonomous drone racing competition being held during the IROS 2016 international conference [2].

To be successful in multi-player drone racing, a player must not only follow the race track in an optimal way, but also compete with other drones through strategic blocking, faking, and opportunistic passing while avoiding collisions. Since unveiling one's own strategy to the adversaries is not desirable, this requires each player to independently predict the other players' future actions. Nash equilibria are a powerful tool to model this and similar multi-agent coordination problems in which the absence of communication impedes full coordination between the agents. In this paper, we propose a novel receding horizon planning algorithm that, exploiting sensitivity analysis within an iterated best response computational scheme, can approximate Nash equilibria in real time. We also describe a vision-based pipeline that allows each player to estimate its opponent's relative position. We demonstrate that our solution effectively competes against alternative strategies in a large number of drone racing simulations. Hardware experiments with onboard vision sensing prove the practicality of our strategy. Video of the experiments: . Drone racing has recently become a popular sport with international competitions being held regularly and attracting a growing public . In these races, human pilots directly control the UAVs through a radio transmitter while receiving a first-person-view live stream from an onboard camera. Human racers need years of training to master the advanced navigation and control skills that are required to be successful in this sport. Many of these skills would certainly prove useful for a robot to safely and quickly move through a cluttered environment in, for example, a disaster response scenario. For this reason, drone racing has attracted a significant interest from the scientific community, which led to the first autonomous drone racing competition being held during the IROS 2016 international conference [2].

Hearing about the violent conditioning that do on a diurnal base around the world is relatively inviting. particular safety and social stability are seriously hovered by the violent conditioning. A variety of styles have been tried to check the violent conditioning which includes installing of surveillance systems. It'll be of great significance if the surveillance systems can automatically descry violent conditioning and give warning or alert signals. The whole system can be enforced with a sequence of procedures. originally, the system has to identify the presence of mortal beings in a videotape frame. also, the frames which are prognosticated to contain violent conditioning has to be uprooted. The in-applicable frames are to be dropped at this stage. Eventually, the trained model detects violence and these frames are independently saved as images. These images are enhanced to descry faces of people involved in the exertion, if possible. The enhanced images along with other necessary details similar as time and position is transferred as an alert to the concerned authority. The proposed system is a deep literacy grounded automatic discovery approach that uses Convolutional Neural Network to descry violence present in a videotape. But, the disadvantage of using just CNN is that, it requires a lot of time for calculation and is less accurate. Hence, a pre-trained model, MobileNetV2, which provides advanced delicacy and acts as a starting point for the structure of the entire model. An alert communication is given to the concerned authorities using telegram operation.

Hearing about the violent conditioning that do on a diurnal base around the world is relatively inviting. particular safety and social stability are seriously hovered by the violent conditioning. A variety of styles have been tried to check the violent conditioning which includes installing of surveillance systems. It'll be of great significance if the surveillance systems can automatically descry violent conditioning and give warning or alert signals. The whole system can be enforced with a sequence of procedures. originally, the system has to identify the presence of mortal beings in a videotape frame. also, the frames which are prognosticated to contain violent conditioning has to be uprooted. The in-applicable frames are to be dropped at this stage. Eventually, the trained model detects violence and these frames are independently saved as images. These images are enhanced to descry faces of people involved in the exertion, if possible. The enhanced images along with other necessary details similar as time and position is transferred as an alert to the concerned authority. The proposed system is a deep literacy grounded automatic discovery approach that uses Convolutional Neural Network to descry violence present in a videotape. But, the disadvantage of using just CNN is that, it requires a lot of time for calculation and is less accurate. Hence, a pre-trained model, MobileNetV2, which provides advanced delicacy and acts as a starting point for the structure of the entire model. An alert communication is given to the concerned authorities using telegram operation.

The number of resident space objects (RSOs) in orbit has increased dramatically within the last ten years. While RSOs pose a serious challenge to our continued use of space, these objects also provide an opportunity to improve on-orbit state estimation. Spacecraft star trackers are commonly used to determine the orientation of its host spacecraft but these sensors are also capable of detecting RSOs. These detections contain orbital positional information that traditional star-images lack and could allow star trackers to provide both position and attitude state estimates. In order for RSO-based optical navigation to be commercially viable, a reliable filter covariance estimate is required. This paper introduces an Unscented Kalman Filter (UKF) for estimating an observing spacecraft's position and attitude based on RSO observations. To ensure this filter is reliable, a technique for bounding the estimate error with the moving standard deviation of its error and the square root of its covariance is introduced. This method provides strong indicator for the reliability of an estimate.

Recently target driven visual navigation strategies have gained a lot of popularity in the computer vision and reinforcement learning community. Unfortunately, most of the current research tends to incorporate sensory input into a reward-based learning approach, with the hope that a robot can implicitly learn its optimal actions through recursive trials. These methods seldom generalize across domains as they fail to exploit natural environment object relationships. We present Memory-utilized Joint hierarchical Object Learning for Navigation in Indoor Rooms (MJOLNIR), a target-driven visual navigation algorithm, which considers the inherent relationship between "target" objects, along with the more salient "parent" objects occurring in its surrounding. Extensive experiments conducted across multiple environment settings show $\approx \textbf{30 %}$ improvement over the existing state-of-the-art navigation methods in terms of the success rate. We also show that our...

This paper aims to explore the use of Visual Inertial Odometry (VIO) for tracking and measurement. The evolution of VIO is first discussed, followed by the overview of monocular Visual Odometry (VO) and the Inertial Measurement Unit (IMU). Next, the related measurement approaches and the use of VIO for measurement have also been discussed. Visual Inertial Odometry is the combination of IMU in the VO system in which the visual information and inertial measurements are combined to achieve an accurate measurement. The algorithm of VO system contains four components, which are camera calibration algorithm, the feature tracker algorithm (usually the KLT algorithm), the rigid motion estimation algorithm, and the algorithm that matches a description of the features points (typically RANSAC algorithm). The IMU is the combination of accelerometer, gyroscopes and magnetometer that measures the linear and angular motion. To fuse the visual and inertial measurements data, there are two differen...

Accurate row detection is crucial for autonomous navigation systems in agricultural applications. Since the crops are typically planted in evenly spaced, parallel rows, crop rows mathematically correlate to a specific spectral signature within the frequency domain. In this study, we propose a robust frequency domain-based image processing pipeline for accurately locating crop rows within images. Using the two-dimensional discrete Fourier transform (DFT), we identify the spacing and direction of crop rows from significant frequency components, and their positions from the phase. To enhance precision, we apply frequency domain interpolation, a Hanning window, and discrete-time Fourier transform (DTFT) to accurately locate peak responses and corresponding phases in the spectrum. Extensive experiments were conducted to validate the accuracy and efficiency of our algorithm. The algorithm accurately detected crop rows in images and estimated lateral position and heading deviations relative to row centerlines, across different plant growth stages and illumination conditions. Additionally, by integrating a linear quadratic Gaussian (LQG) navigation controller with this vision-based crop row detection algorithm, we successfully enabled a robot to follow crop row centerlines with a mean absolute tracking error of 3.74 cm. These results highlight the potential of our frequency domain-based approach for enhancing precision agriculture through robust crop row detection and navigation.

This paper proposes a set of practical extensions to the vision-based Monte Carlo localization for RoboCup Sony AIBO legged robot soccer domain. The main disadvantage of AIBO robots is that they have a narrow field of view so the number of landmarks seen in one frame is usually not enough for geometric calculation. MCL methods have been shown to be accurate and robust in legged robot soccer domain but there are some practical issues that should be handled in order to maintain stability/elasticity ratio in a reasonable level. In other words, the fast convergence ability is required in case of kidnapping. But on the other hand, fast converge can be vulnerable when an occasional bad sensor reading is received. In this work, we presented four practical extensions in which two of them are novel approaches and the remaining ones are different from the previous implementations.

Changes resulting from the publishing process, such as peer review, editing, corrections, structural formatting, and other quality control mechanisms may not be reflected in this document. Changes may have been made to this work since it was submitted for publication. The full version of the article is available on Elsevier's sites with

Unmanned Aerial Vehicles (UAVs) are increasingly popular and widely used across various sectors, including industry, academia, and the military (Mohsan, Khan, Noor, Ullah & Alsharif, 2022). They play a crucial role in global defense, particularly in hostile environments, as they typically require fewer personnel for operations and maintenance than conventional aircraft (Boyle, 2012). In Portugal, a coastal state, UAVs hold significant potential for monitoring and surveillance tasks, even though their implementation has not yet occurred. They can provide continuous coverage at a low cost, making them valuable assets for search and rescue operations and mission coordination. Automating their processes becomes essential as the number of UAVs in operation rises. This automation will help manage more UAVs with fewer personnel while ensuring reliability throughout all operational phases, particularly during the most challenging phases: take-off and landing (Boyle, 2012). Most UAV accidents occur during these operations, mainly due to human factors (Grindley et al., 2024), and automating these operations is essential to improving the overall reliability of the UAV system. Our research aims to make a significant contribution to enhancing UAV operations. We will be using multirotor UAVs, which make take-off straightforward. Consequently, our primary focus will be enhancing the landing operation to improve the system's reliability.

Landing on Mars or other planetary bodies, close to a predetermined target landing spot, in an area of rough terrain, is a difficult and risky task. Accurate navigation relative to the planetary surface is necessary, together with the detection of possible hazards like boulders or steep slopes. This is being made possible by developments in vision-based guidance techniques and on-board processing technology. This paper describes a novel EKF to estimate the position and the attitude quaternion of a planetary lander during the approach phase from high-gate to low-gate, using information from vision integrated with inertial data and other GNC sensor data.

The launchers GNC performance is strictly depending upon the knowledge of the system properties and its evolution throughout the flight. The GNC design and verification integrates mathematical models derived at sub-system level to predict system level behaviour. Their improvement is a key task through the development and validation of a new launch vehicle. Hence the exploitation of flight data shall be maximized in order to reduce at minimum the lack of accuracy in the mathematical models. It is clear that the exploitation of in-flight data during qualification flights in order to improve the launch vehicle models plays a key role in the tuning of the GNC system algorithms and architecture for future flights. This paper presents the Model Validation Framework for Launchers which is a set of tools and algorithms for the determination and validation of accurate mathematical models and parameters based on pre-, inand post-flight measurement data. The Model Validation Framework for Laun...

Missions to small celestial bodies rely heavily on optical feature tracking for characterization of and relative navigation around the target body. While deep learning has led to great advancements in feature detection and description, training and validating data-driven models for space applications is challenging due to the limited availability of large-scale, annotated datasets. This paper introduces AstroVision, a large-scale dataset comprised of 115,970 densely annotated, real images of 16 different small bodies captured during past and ongoing missions. We leverage AstroVision to develop a set of standardized benchmarks and conduct an exhaustive evaluation of both handcrafted and data-driven feature detection and description methods. Next, we employ AstroVision for end-to-end training of a state-of-the-art, deep feature detection and description network and demonstrate improved performance on multiple benchmarks. The full benchmarking pipeline and the dataset will be made publicly available to facilitate the advancement of computer vision algorithms for space applications.

In aviation applications, navigation integrity is paramount. Integrity of GPS systems is well established with set standards. Vision-based navigation systems have been found to be an adequate substitute for GPS when it is unavailable but are unlikely to be utilized until there is a measure for system integrity. Work has been done to detect the effect of a single measurement pair being corrupted with a bias; however, the measurement geometry varies greatly with the environment. The environment could be sparse in visual features to track, or the environment could be rich with features. With more features, there is a greater probability of having multiple corrupted measurements. It is essential that multiple corrupt measurements are detected and excluded to assure the integrity and reliability of the system. In addition, misalignment errors in the camera system results in systematic errors that are undetectable by current methods. This dissertation focuses on understanding the existing integrity monitoring methods and using them for the detection of multiple errors in vision-based navigation systems, as well as, developing a technique for detecting systematic errors due to camera misalignment and scaling. These methods are developed analytically and verified by using simulations. These simulations serve to demonstrate the usefulness of these methods in achieving the goal of this research to further the area of integrity monitoring for vision systems, so it could eventually be used as a trusted system and as a backup for GPS navigation.

Smart and precision agriculture seeks to boost the efficiency of operations and crop yield by using modern technology. Modern tools such as sensors, imagery cameras, and deep learning enable farmers to identify and control weeds, pests, and diseases in real-time. A robotic platform can carry these modern types of equipment and achieve the mentioned objectives precisely. Automatic and accurate navigation of this autonomous robot in agricultural fields is essential for performing these precision tasks. An agricultural robotic platform was designed and developed for row crop fields. The robot navigation system comprises two main components: a vision-based row detection system for path tracking and a motion controller system. The vision-based guidance system processes acquired image data from a tilted camera in front of the robot to identify the crop row's position. The Hough transform method was used to determine the position of the crop rows. Using the resultant guidance line equations, the motion controller directs the robot to move automatically between rows without harming the crops. Differential speed steering allows both wheels on the robot to rotate at different speeds. The steering system improved the robot position error by controlling both powered wheel speeds. To move the robot among the crop rows, it generates the wheel speed difference command. The robotic platform effectively followed the rows of sugar beets at a velocity of 0.5 m/s, exhibiting an average lateral offset of 12 mm and a standard deviation of 22 mm.

In this article, a robot formation control strategy based on a vision-based follow-the-leader scenario is proposed, with emphasize on its reliability. On the one hand, perception is enhanced by the control of a motorized zoom. On the other hand, bidirectional and non-oblivious [1] control is implemented, with an odometry-based fault detection of vision-based information and a leader path-planning strategy to improve its visibility. After introduction of the control strategy, extensive experimental results are presented.

State of the practice in navigation toward and around small celestial bodies heavily relies on ground support and human skill, in particular for perception-based operations such as optical navigation and mapping. This leads to longer and more complex mission operations and subsequently higher cost. Furthermore, it imposes limitations for certain missions such as fast fly-bys or multi-agent operations. In this work, we present an autonomous navigation strategy for approaching unexplored small bodies. During the approach, we estimate the body’s physical properties as well as the spacecraft relative trajectory and associated uncertainties. This strategy, which is solely based on optical measurements, begins as soon as the body becomes resolved in the navigation camera and terminates at the start of proximity operations, when the spacecraft makes its first trajectory correction to stay in the vicinity of the body. Our multi-phased approach uses light curve analysis for estimating the bo...

The goal of this project is to develop a new method of Route following for mobile robots in a GNSSdenied environment like urban canyons or indoor. We used a robust biologically constrained neural model inspired by ants developed previously in simulation to assess the familiarity index of a panorama. A visual compass algorithm consists in determining the orientation of the maximum familiarity index with respect to the learned panoramas along a path. A car-like robot was equipped with a 220°fisheye camera. The visual compass algorithm used low resolution images of 44x44 pixels (5°/pixel) indoors and outdoors to determine the direction to follow the previously visually-learned path. Finally, the car-like robot was automated to recall the learned path indoors. The biologically constrained neural model compressed the visual information with a high efficiency so that the visual memory has a very low footprint of a few tens of kilobits that does not depend directly on the path length.

Sensorimotor actions are thus often derived from high-level, quantifiable, optimality principles assigned to each task, utilizing consolidated tools in optimal control theory. The planned actions are derived on the ground and transferred onboard where controllers have the task of tracking the uploaded guidance profile. Here we argue that end-toend neural guidance and control architectures (here called G&CNets) allow transferring onboard the burden of acting upon these optimality principles. In this way, the sensor information is transformed in real time into optimal plans thus increasing the mission autonomy and robustness. We discuss the main results obtained in training such neural architectures in simulation for interplanetary transfers, landings and close proximity operations, highlight-1

Autonomous navigation of generic monocular quadcopter in the natural environment requires sophisticated mechanism for perception, planning and control. In this work, we have described a framework which performs perception using monocular camera and generates minimum time collision free trajectory and control for any commercial quadcopter flying through cluttered unknown environment. The proposed framework first utilizes supervised learning approach to estimate the dense depth map for video stream obtained from frontal monocular camera. This depth map is initially transformed into Ego Dynamic Space and subsequently, is used for computing locally traversable way-points utilizing binary integer programming methodology. Finally, trajectory planning and control module employs a convex programming technique to generate collision-free trajectory which follows these waypoints and produces appropriate control inputs for the quadcopter. These control inputs are computed from the generated trajectory in each update. Hence, they are applicable to achieve closed-loop control similar to model predictive controller. We have demonstrated the applicability of our system in controlled indoors and in unstructured natural outdoors environment.

IstiABot is a mobile robot made for education and research purposes. This robot aims to be modular, easy to modify and used by first year students as well as last year students and researchers. To achieve those requirements, the robot is built on the top of a CANBus Network. This paper presents the robot and the approach behind its building. It also presents an educational application of the platform (tuning a PID controller) and a research application (Simultaneous Localization And Mapping-SLAM-experimentations). Finally it concludes about this experience and introduce two other robots that were built based on the IstiABot. As it was wanted for the robot to be as free as possible, all the source code and 3D modeling are available.

Augmented reality (AR) refers to seamlessly inserted virtual objects into the real world in a real-time way. The real-time requirement can be associated with pose estimation or, equivalently, camera pose localization. Herein, we provide an overview of the camera pose localization domain for AR, explain the pose estimation problem, and provide a survey of relevant image-based localization methods. We highlight the localization problem via feature extraction and matching through mapping the scene and constructing the 3D scene structure.

The performance of visual odometry is dependent upon the quality of features selected for computing the frame-to-frame transformation. In order to ensure the quality of selected features, conventional approaches consider the spatial distribution of the selected features, in addition to their counts and matching scores, in which a small number of features are selected randomly from each of the uniformly distributed buckets. In this paper, we show that features can be selected optimally, rather than randomly, using a welldefined mathematical formalism. The proposed method of optimal feature selection minimizes the degree of uncertainty in estimating the essential, fundamental, or homography matrix involved in visual odometry by maximizing the orthogonality index of individual equations and constraints associated with computation. We found that, at a constant noise level, the mean of the residual error and the variance of an estimated essential, fundamental, or homography matrix decrease monotonically with increasing orthogonality index. The simulation validates the increased accuracy of the feature selection based on the proposed orthogonality index compared with the conventional random selection. For instance, it enhances accuracy by as much as 35% when a small number of feature sets, say, 20 feature sets, are used. The experiments using the KITTI and Devon Island datasets further reinforce the performance enhancement of simulations by 9% and 20%, respectively.

After arriving at the near-Earth asteroid (101955) Bennu in late 2018, the OSIRIS-REx spacecraft will execute a series of observation campaigns and orbit phases to accurately characterize Bennu and ultimately collect a sample of pristine regolith from it's surface. While in the vicinity of Bennu, the OSIRIS-REx navigation team will rely on a combination of ground-based radiometric tracking data and optical navigation (OpNav) images to generate and deliver precision orbit determination products. Long before arrival at Bennu, the navigation team is performing multiple orbit determination simulations and thread tests to verify navigation performance and ensure interfaces between multiple software suites function properly. In this paper, we summarize the results of an independent orbit determination simulation of the Orbit B phase of the mission performed to test the interface between the OpNav image processing and orbit determination software packages.

Sensorimotor actions are thus often derived from high-level, quantifiable, optimality principles assigned to each task, utilizing consolidated tools in optimal control theory. The planned actions are derived on the ground and transferred onboard where controllers have the task of tracking the uploaded guidance profile. Here we argue that end-toend neural guidance and control architectures (here called G&CNets) allow transferring onboard the burden of acting upon these optimality principles. In this way, the sensor information is transformed in real time into optimal plans thus increasing the mission autonomy and robustness. We discuss the main results obtained in training such neural architectures in simulation for interplanetary transfers, landings and close proximity operations, highlight-1

In this paper, we propose a robust and integrated visual odometry framework exploiting the optical flow and feature point method that achieves faster pose estimate and considerable accuracy and robustness during the odometry process. Our method utilizes optical flow tracking to accelerate the feature point matching process. In the odometry, two visual odometry methods are used: global feature point method and local feature point method. When there is good optical flow tracking and enough key points optical flow tracking matching is successful, the local feature point method utilizes prior information from the optical flow to estimate relative pose transformation information. In cases where there is poor optical flow tracking and only a small number of key points successfully match, the feature point method with a filtering mechanism is used for posing estimation. By coupling and correlating the two aforementioned methods, this visual odometry greatly accelerates the computation time...

The article presents an approach to the control of a UAV on the basis of 3D landmark observations. The novelty of the work is the usage of the 3D RANSAC algorithm developed on the basis of the landmarks' position prediction with the aid of a modified Kalman-type filter. Modification of the filter based on the pseudo-measurements approach permits obtaining unbiased UAV position estimation with quadratic error characteristics. Modeling of UAV flight on the basis of the suggested algorithm shows good performance, even under significant external perturbations.

The goal of this project is to develop a new method of Route following for mobile robots in a GNSSdenied environment like urban canyons or indoor. We used a robust biologically constrained neural model inspired by ants developed previously in simulation to assess the familiarity index of a panorama. A visual compass algorithm consists in determining the orientation of the maximum familiarity index with respect to the learned panoramas along a path. A car-like robot was equipped with a 220°fisheye camera. The visual compass algorithm used low resolution images of 44x44 pixels (5°/pixel) indoors and outdoors to determine the direction to follow the previously visually-learned path. Finally, the car-like robot was automated to recall the learned path indoors. The biologically constrained neural model compressed the visual information with a high efficiency so that the visual memory has a very low footprint of a few tens of kilobits that does not depend directly on the path length.

An innovative approach for navigation in non-GPS environment is presented based on all source adaptive fusion of any available information encompassing passive imaging data, digital elevation terrain data, IMU/GPS, altimeters, and star tracker. The approach provides continuous navigation through non-GPS environment and yields an improved navigation in the presence of GPS. The approach also provides reduced target location error and moving target indication.

Guest Editorial This special issue of the Journal of Networks includes extended versions of extended versions of selected best papers accepted and presented at the 8th International Joint Conference on e-Business and Telecommunications (ICETE 2011). Eight papers were selected based on their excellent review scores. Their authors were invited to submit extended versions, which have undergone a second review process to guarantee that all the papers included in this Special Issue have been rigorously peer-reviewed. The accepted papers cover areas like empirical evaluations, simulation modeling and theoretical studies addressing a variety of topics related to security in telecommunications, wireless networking, and localization. The first paper "Simplified Scheduling for Underwater Acoustic Networks" is authored by W. Van Kleunen, N. Meratnia and P. J. M. Havinga. It proposes an interesting solution to the scheduling and routing issues in underwater acoustic networks, giving a simplified set of constrains. These constrains are applied to three new scheduling algorithms: centralized for maximum throughput, distributed for low computational costs and centralized for low-latency end-toend communications. The evaluation by simulation explores extensively the benefits obtained when the new set of constrains is applied to the proposed algorithms. The second paper, entitled "Power Management Mechanism Exploiting Network and Video Information over Wireless Links" authored by C. Bouras, S. Charalambides, K. Stamos, S. Stroumpis and G. Zaoudis, describes some cross-layer mechanisms to manage power consumption in wireless communications. The authors focus their analysis on video transmissions. The scenarios proposed are simulated using ns-2 simulator, obtaining relevant results referred to the video quality and the associated power consumption. In the third paper, "QoS-Aware Multipath Communications over MANETs", by M. Obaidat, M. Ali, I. Shahwan, M. S. Obaidat, and S. Obeidat, the authors propose a QoS aware routing protocol based on multipath routing. They try to minimize the packet delay providing more than one path between source and destination, and choosing the best path based on the quality of the link. The new protocol is compared against another well-known single path routing protocol (AODV) by means of simulation over OPNET. The results show that the new routing protocol outperforms AODV in terms of end-to-end delay, but at the cost of a greater protocol overhead. The fourth paper, "Fingerprint Indoor Position System Based On Bitcloud and Openmac", authored by J. A. Gómez, A. V. Medina, E. Dorronzoro, O. Rivera and M. Merino, uses fingerprints as an alternative approach to estimate location of mobile nodes in WSNs. The fingerprints are used in two algorithms, being the second one (centroid) the algorithm that offers more accuracy. The algorithms are tested in two real implementations (bitcloud and openmac) over a specific scenario. The results show that openmac can overtake the problems detected in bitcloud and improve the precision on localization. The fifth paper is entitled "Localization Method For Low Power Wireless Sensor Networks", authored by D. F. Larios, J. Barbancho, F. J. Molina and C. León. This paper proposes a range-free localization algorithm that uses fuzzy logic to process RSSI and estimate the position of mobile devices. This method causes less localization errors than other well-known localization methods and, at the same time, it improves power consumption of the mobile nodes. The sixth paper, "Development Tools For Context Aware And Secure Pervasive Computing in Embedded Systems (PECES) Middleware", authored by R. Zhao, K. Selvarajah and N. Speirs presents a set of tools to enable the communication among heterogeneous devices across multiple smart devices and platforms in a secure manner. PECES and its tools are especially useful for experienced developers of services and applications that desire a clean deployment of their applications regardless of the platform for which they are developing. PECES has been evaluated by means of heuristic techniques that conclude that the majority of interviewed people assert that PECES is useful, and easy to use. The seventh paper "Performance of OpenDPI in Identifying Sampled Network Traffic", authored by J. Khalife, A. Hajjar and J. Diaz, studies the impact of sampling traffic on the performance of OpenDPI, by means of two different ways of gathering information: per-packet sampling and per-flow sampling, in order to determine the reduction of the input data and the accuracy of the classification. Results show that the per-flow packet is more convenient for sampling; showing very high accuracy rates. The last paper entitled, "Anomaly Detection Using Metaheuristic Firefly Harmonic Clustering", authored by M. H. A. C. Adaniya, T. Abrão and M. Lemes Proença presents a clustering algorithm based on K-Harmonic means (KHM) and Firefly Algorithm (FA) for the detection of network volume anomalies due to device failing or misconfiguration. As a consequence of the application to the real data collected from a proxy server, and a further statistic analysis, the algorithm achieves about 80% of true positive recognitions and 20% of false positive recognitions, demonstrating the satisfactory behaviour of this algorithm. The guest editors would like to thank all the authors and reviewers for their valuable contributions to this special issue. We also thank the editorial staff of Journal of Networking for their support. We hope that the papers selected in this special issue will become useful resources for researchers and practitioners in these areas.

This paper addresses the problem of vision-based navigation and proposes an original control law to perform the navigation. The overall approach is based on an appearance-based representation of the environment, where the scene is directly defined in the sensor space, by a database of images acquired during a learning space. Within this context, a path to perform is described by a set of images, or image path extracted from the database. This image path is designed so that it provides enough information to control the robotic system. The central point of this paper is the closed-loop control law that drives the robot to its desired position using this image path. This control does not require neither a global 3D reconstruction, nor a temporal planning step. Furthermore, the robot is not constrqined to converge directly upon each image of the path but chooses automatically its trajectory. We propose a qualitative visual servoing, enabling to enlarge the convergence space towards an interval of confident position. We propose and use specific visual features which ensure that the robot navigates within the visibility path. Experimental simulations are given to show the effectiveness of this method for controlling the motion of a camera in three-dimensional environments (free-flying camera, or camera moving on a plane). Furthermore, experiments realized with a robotic arm observing a planar scene are also presented.

RoboSantral, An autonomous mobile robot which has been designed and realized in order to guide the visitors through a university campus, is presented in this paper. This robot accompanies guests through the campus and gives presentations on predefined locations. Location data is obtained from GPS sensors. Targets such as faculty buildings, museums etc... are recognized by the image processing of pre-defined tags. As microprocessor and microcontroller, Raspberry Pi and Arduino are used respectively.

This paper presents a framework for tracking a mobile ground target (MGT) using a fixed-wing unmanned aerial vehicle (UAV). Challenges from pure theories to practical applications, including varying illumination, computational limits and a lack of clarity are considered. The procedure consists of four steps, namely: target detection, target localization, states estimation and UAV guidance. Firstly, the MGT in the wild is separated from the background using a Laplacian operator-based method. Next, the MGT is located by performing coordinate transformations with the assumption that the altitude of the ground is invariant and known. Afterwards, a Kalman filter is used to estimate the location and velocity of the MGT. Finally, a modified guidance law is developed to guide the UAV to circle and track the MGT. The performance of our framework is validated by simulations and a number of actual flight tests. The results indicate that the framework is effective and of low computational complexity, and in particular our modified guidance law can reduce the error of the tracking distance by about 75% in specified situations. With the proposed framework, such challenges caused by the actual system can be tackled effectively, and the fixed-wing UAV can track the MGT stably.

This paper presents a framework for tracking a mobile ground target (MGT) using a fixed-wing unmanned aerial vehicle (UAV). Challenges from pure theories to practical applications, including varying illumination, computational limits and a lack of clarity are considered. The procedure consists of four steps, namely: target detection, target localization, states estimation and UAV guidance. Firstly, the MGT in the wild is separated from the background using a Laplacian operator-based method. Next, the MGT is located by performing coordinate transformations with the assumption that the altitude of the ground is invariant and known. Afterwards, a Kalman filter is used to estimate the location and velocity of the MGT. Finally, a modified guidance law is developed to guide the UAV to circle and track the MGT. The performance of our framework is validated by simulations and a number of actual flight tests. The results indicate that the framework is effective and of low computational compl...

This research focuses on developing a helicopter autonomous ship landing algorithm based on the real helicopter ship landing procedure which is already proven and currently used by Navy pilots. It encompasses the entire ship landing procedure from approach to landing using a pilotinspired vision-based navigation system. The present thesis focuses on the first step towards achieving this overarching objective, which involves modeling the flight dynamics and control of a helicopter and some preliminary simulations of a UH-60 (Blackhawk) helicopter landing on a ship. I owe my deepest thanks to my family-my father, Byungkyun Lee, my mother, Okhwa Lee, and my sister Younhye Lee, who have always stood by me and supported me through my career. My wife, Sunhye Lee has sacrificed many things in her life and beared with my busy life. I promise that I will pay back my debts in the rest of life. My little one, Jaeho Lee is my reason for life. He motivated me to become a father he can be proud of. I always love you as you are.

The exploration of celestial bodies such as the Moon, Mars, or even smaller ones such as comets and asteroids, is the next frontier of space exploration. One of the most interesting and attractive purposes from the scientific point of view in this field, is the capability for a spacecraft to land on such bodies. Monocular cameras are widely adopted to perform this task due to their low cost and system complexity. Nevertheless, image-based algorithms for motion estimation range across different scales of complexities and computational loads. In this paper, a method to perform relative (or local) terrain navigation using frame-to-frame features correspondences and altimeter measurements is presented. The proposed image-based approach relies on the implementation of the implicit extended Kalman filter, which works using nonlinear dynamic models and corrections from measurements that are implicit functions of the state variables. In particular, here, the epipolar constraint, which is a ...

This paper presents a new Strap-down Inertial Navigation System/Spectrum Red-Shift/Star Sensor (SINS/SRS/SS) system integration methodology to improve the autonomy and reliability of spacecraft navigation using the spectrum red-shift information from natural celestial bodies such as the Sun, Jupiter and the Earth. The system models for SINS/SRS/SS integration are established. The information fusion of SINS/SRS/SS integration is designed as the structure of the federated Kalman filter to fuse the local estimations of SINS/SRS and SINS/SS integrated subsystems to generate the global state estimation for spacecraft navigation. A new robust adaptive unscented particle filter is also developed to obtain the local state estimations of SINS/SRS and SINS/SS integrated subsystems in a parallel manner. The simulation results demonstrate that the proposed methodology for SINS/SRS/SS integration can effectively calculate navigation solutions, leading to strong autonomy and high reliability for spacecraft navigation.

The article presents an overview of the theoretical and experimental work related to unmanned aerial vehicles (UAVs) motion parameters estimation based on the integration of video measurements obtained by the on-board optoelectronic camera and data from the UAV's own inertial navigation system (INS). The use of various approaches described in the literature which show good characteristics in computer simulations or in fairly simple conditions close to laboratory ones demonstrates the sufficient complexity of the problems associated with adaption of camera parameters to the changing conditions of a real flight. In our experiments, we used computer simulation methods applying them to the real images and processing methods of videos obtained during real flights. For example, it was noted that the use of images that are very different in scale and in the aspect angle from the observed images in flight makes it very difficult to use the methodology of singular points. At the same tim...

UAV's (Unmanned Aerial Vehicle) in general fly in open space, far from any obstacles and relay on external beacons, mainly GPS (Global Positioning System) to localize them and navigate. This approach precludes them from evolving autonomously at low altitudes, in cluttered and confined environments. The autonomous flight in these cluttered environments like houses or urban canyons requires high maneuverability, fast mapping from sensors to actuators and very limited overall system weight. The flying animals are well capable of coping with such situations and so a much similar navigation method is to be adopted. The optical flow navigation will be best suited for these situations. The optical flow is the pattern of apparent motion of objects, surfaces, and edges in visual scene caused by the relative motion between the observer and scene. The use of hardware and power consumption in the optical flow navigation is very much less when compared to other navigational aids and thus it ...

This paper presents a real-time system for ball detection and tracking system which is reliable in any conditions. Images from the webcam are processed by openCV library running on a Raspberry Pi to move the camera pan and tilt servo and two DC motors to drive the robot body using the Arduino Nano microcontroller.  The webcam is integrated in a robot prototype to represent the wheel football robot type. The results show that a ball tracking webcam system is obtained with the capability to detect a ball with a diameter of 17cm within a maximum distance of 200 cm, a stable ball reading when the light intensity is at 32 lux and above. Furthermore, the experimental results demonstrated the system’s robustness in detecting and tracking ball in different distance and ligthing conditions.