A multi-sensor fusion-based UAV autonomous localization system: design and implementation for complex terrain mapping

For navigation, perception, control, and all-around safe operation, autonomous vehicles ought to know its accurate location and orientation in any kind of weather and traffic. Because autonomous vehicles operate under a wide range of conditions, several sensors are utilized to assure the necessary performance. The Extended Kalman filter (EKF) and Unscented Kalman filter (UKF) are utilized as sensor fusion algorithms to fuse low-cost INS/GNSS/Odometer. Nevertheless, there are a number of sources of error exist. It must be limited by the acceptable amount of risk for the application. Therefore, in this paper, we focus on Horizontal Protection Level (HPL) and choose the kSigma algorithm to compute it. By using tightly coupled integration, the comparison between EKF and UKF in the sense of HPL is analyzed. First, we analyzed the difference RMS position error between EKF and UKF. Then, the result both algorithms in Horizontal Position Error (HPE) subject to HPL are investigated. Overall, the UKF solution has a better result in accuracy and integrity than EKF. It indicated with lower percentage of epoch UKF solution that satisfied HPE < HPL condition than EKF and lower RMS position error.

Highly robust and accurate multi-sensor fusion localization system for complex and challenging scenarios

Currently, autonomous mining vehicles are using GNSS for localization. Due to atmospheric phenomena, the GNSS signal becomes unstable, making autonomous equipment stop their movements, thus decreasing the mine's productivity. This paper presents a method to estimate the 2D localization of ground vehicles through the optical flow from images of a camera pointed at the ground, IMU, and wheel encoder, focusing on mining environments. Using a ground-facing camera is more robust to particulates in the air, like fog and dust, than techniques using horizon-facing sensors. We analyze five implementations for localization: (1) using wheel encoders, (2) a visual-only method, (3) using the IMU orientation and linear displacement by visual information, (4) obtained by merging wheel encoder and IMU data using Extended Kalman Filter (EKF), and (5) an EKF using visual, encoder, and IMU data. We perform tests in mining-like environments in simulation and field experiments. Simulations are implemented in CoppeliaSim software and make use of realistic textures. In the field experiments, we use a mobile robot equipped with a camera, IMU, and GNSS receiver with RTK correction, which we consider the robot's actual position (ground truth). Results show that the proposed methods are promising but need to become more accurate for use in heavy mining vehicles.

In this paper, the performance of different hybrid navigation filters exploiting GPS, Galileo and 5G Time Of Arrival (TOA) measurements in multipath environment are studied. For the realism of the study, realistic propagation channels must be considered and their impacts on the received signals processing must be accurately modelled. GNSS signal mathematical models in multipath environment have been analyzed for a long time. However, 5G mathematical models in a realistic multipath environment are still in its early stages of analysis. This article is divided in three main parts. The first part is dedicated to the identification of compliant GNSS and 5G signal propagation channel models; SCHUN is selected for GNSS and QuaDRiGa is selected for 5G. Based on this, the correlator output mathematical models for 5G signals and GNSS signals are derived. The second part tackles the accurate characterization of the pseudo range errors due to propagation channels shadowing and multipath effect as well as thermal noise. This step is required for the correct derivation of the navigation filters. Indeed, the study will focus on Extended Kalman Filters (EKF) and Unscented Kalman Filters (UKF); both assume a Gaussian distribution of the errors. Therefore, by optimally characterizing the errors, the performances of the filters are expected to be improved. The last part consists in validating through simulations the theory and mathematical models developed in the first and second parts.

This paper proposes a novel motion planning method for an autonomous ground mobile robot to address dynamic surroundings, nonlinear program, and robust optimization problems. A planner based on the recurrent fuzzy neural network (RFNN) is designed to program trajectory and motion of mobile robots to reach target. And, obstacle avoidance is achieved. In RFNN, inference capability of fuzzy logic and learning capability of neural network are combined to improve nonlinear programming performance. A recurrent frame with self-feedback loops in RFNN enhances stability and robustness of the structure. The extended Kalman filter (EKF) is designed to train weights of RFNN considering the kinematic constraint of autonomous mobile robots as well as target and obstacle constraints. EKF's characteristics of fast convergence and little limit in training data make it suitable to train the weights in real time. Convergence of the training process is also analyzed in this paper. Optimization technique and update strategy are designed to improve the robust optimization of a system in dynamic surroundings. Simulation experiment and hardware experiment are implemented to prove the effectiveness of the proposed method. Hardware experiment is carried out on a tracked mobile robot. An omnidirectional vision is used to locate the robot in the surroundings. Forecast improvement of the proposed method is then discussed at the end.

In underwater localization systems several sources of error may impact in different ways the accuracy of the final position estimates. Through simulations and statistical analysis it is possible to identify and characterize such sources of error and their relative importance. This is especially of use when an accurate localization system has to be designed within required accuracy prescriptions. This approach allows one to also investigate how much these sources of error influence the final position estimates achieved by an Extended Kalman Filter (EKF). This paper presents the results of experiments designed in a virtual environment used to simulate real acoustic underwater localization systems. The paper intends to analyze the main parameters that significantly influence the position estimates achieved by a Short Baseline (SBL) system. Specifically, the results of this analysis are presented for a proprietary localization system constituted by a surface platform equipped with four acoustic transducers used for the localization of an underwater target. The simulator here presented has the purpose of simulating the hardware system and modifying some of its design parameters, such as the base-line length and the errors on the GPS and Inertial Measurement Unit (IMU) units, in order to understand which parameters have to modify for improving the accuracy of the entire positioning system. It is shown that statistical analysis techniques can be of help in determining the best values of these parameters that permit to improve the performance of a real hardware system.

This paper presents design considerations of a GPS-aided localization system for unmanned vehicles used in outdoor applications. The system proposed in this paper is based on Extended Kalman Filter (EKF) and also integrates Global Positioning System (GPS) measurements. Localization and navigation systems are base components of all unmanned vehicles since they give unmanned vehicles the ability of perceiving the environment in order to localize themselves and to navigate to a target. The advantage of the proposed system over a GPS based localization system is that the system works even if the GPS receiver does not receive any GPS signals. In this study, firstly proposed EKF-based localization system is explained, and then how to integrate GPS measurements into this localization system is explained. With simulation studies in MATLAB, the effectiveness of the system is shown. The simulations show that the proposed localization system gives accurate results with negligible positional errors.

Main goal of navigation-system design is to select a suite of sensors and a computational approach that allows the accuracy, update rate, and bandwidth specifications to be achieved in a cost-effective manner. Today, most of the navigation systems rely on inertial sensors measurements or GPS measurements. There are many difficulties which are present in the INS. Visual information is the must nowadays in terms of navigation and guidance measurements for autonomous aerial and ground vehicles. This visual information can be efficiently used in the Simultaneous Localization and Map building (SLAM) problem. In this paper, we have developed a complete effective airborne SLAM algorithm with INS and Visual systems by implementing an Extended Kalman Filter (EKF). Here we have proposed a solution to the well-known shortcomings that complicate the application of the EKF in large, real-world environments by removing the landmarks from the EKF and creating a special module called map management. We proposed a methodology based on circle intersections and we demonstrate the effectiveness of this methodology along with all the modules of our SLAM algorithm with virtual images taken from a downward looking virtual camera. I. Introduction fficient localization and mapping algorithms are vital for autonomous operations of unmanned systems. For many civilian and military aerial applications, the capabilities of autonomous operation without any or with minimal human intervention are of major importance. Simultaneous localization and map building (SLAM) problem continues to draw considerable attention of the research community due to the advantages it can offer, if efficiently solved, in building autonomous systems. SLAM examines the ability of an autonomous vehicle, starting in an unknown environment, to incrementally build an environment map and simultaneously localize itself within this map. A large number of research work related to SLAM were published in 2, 3, 5, 6, 7, 8 and proposing probabilistic- based solutions able to process noisy measurements (sensing errors) to be turned into accurate map. In this paper, we describe a technique dealing with airborne SLAM, where onboard sensors like INS and vision are used. A use of Extended Kalman Filter (EKF) is proposed without an increase of the number of estimated states that usually pose a problem and complicate the application of the EKF for SLAM in large, real-world environments. Landmarks are not directly included in the EKF estimation process as we created a special module called map management dealing with map measurements. In this module we proposed and implemented methodology what we called it, circle intersection. We demonstrated the effectiveness of the implemented methodology and our SLAM algorithm with virtual images taken from a downward virtual camera system. This paper is organized as follows. In section 2, we detail about the SLAM problem by presenting up to date related work in this active area. Section 3 is reserved for the definition of the Inertial/Vision EKF estimation algorithm. In Section 4 simulation experiments and results are given followed by a short conclusion and future work in section 5.

In order to solve the problem of time-delay in non-linear multi-sensor information fusion, a simulation model of moving target tracking in sensor networks with the time-delay state and observation system is established. Using the augmented matrix to transform the time-delay system into the non-time-delay system, an improved Covariance Intersection (CI) Extended Kalman filter (EKF) data fusion algorithm for multi-sensor systems with time-delay is presented. This method avoids calculating any two local filter error cross-covariance matrices and greatly reduces the computational complexity and time. Analysing the precision of this method, comparing the precision of improved CI EKF data fusion algorithm and the locally optimal fusion EKF algorithm. The results show that the precision of improved CI EKF data fusion algorithm is higher than that of the local EKF and close to the precision of the optimal EKF fusion.

Indoor localization systems using WiFi received signal strength (RSS) or pedestrian dead reckoning (PDR) both have their limitations, such as the RSS fluctuation and the accumulative error of PDR. To exploit their complementary strengths, most existing approaches fuse both systems by a particle filter. However, the particle filter is unsuitable for real time localization on resource-limited smartphones, since it is rather time-consuming and computationally expensive. On the other hand, the light computation fusion approaches including Kalman filter and its variants are inapplicable, since an explicit RSS-location measurement equation and the related noise statistics are unavailable. This paper proposes a novel data fusion framework by using an extended Kalman filter (EKF) to integrate WiFi localization with PDR. To make EKF applicable, we develop a measurement model based on kernel density estimation, which enables accurate WiFi localization and adaptive measurement noise statistics estimation. For the PDR system, we design another EKF based on quaternions for heading estimation by fusing gyroscopes and accelerometers. Experimental results show that the proposed EKF based data fusion approach achieves significant localization accuracy improvement over using WiFi localization or PDR systems alone. Compared with a particle filter, the proposed approach achieves comparable localization accuracy, while it incurs much less computational complexity.

GNSS signals may present anomalies that degrade the positioning performance of GNSS receivers. Signal Quality Monitoring (SQM) is normally used to detect and to characterize these anomalies. This is required for the GNSS operators and integrity services to determine when a satellite should be considered as faulty and draw conclusions about the type of the fault. In this paper, we present a new SQM algorithm that tracks the GNSS signal and possible channel deformations by using a novel methodology based on the Extended Kalman Filter (EKF). The EKF is designed such that the measurement update is performed in post-correlation and using multiple correlators. After the estimation of the channel response, we add a detection step to determine if the channel deviates from the nominal signal transmission scenario (i.e., the single path propagation). Results suggests that the performance of the delay estimation with the proposed EKF structure outperforms the classical Delay-Locked-Loop (DLL) estimation, especially in the presence of distortions. Furthermore, it can reliably detect anomalous signal deformations as specified by ICAO threat model.

Nowadays, the Inertial Navigation System/Global Navigation Satellite System (INS/GNSS) integrated navigation system is widely used in many applications. The extended Kalman Filter (EKF) is a popular data fusion method for the INS/GNSS integrated navigation system. However, the process and measurement noise covariance matrices of the EKF cannot be modelled accurately due to varied scenes and complicated GNSS signal errors in urban environments, which undermines or deteriorates the EKF's performance. To mitigate noise covariance uncertainties' influence, this paper proposes an adaptive EKF algorithm named SARSA EKF, which enables the State-Action-Reward-State-Action (SARSA) method in EKF to realise the autonomous selection of the noise covariance matrices based on the Q-value. Meanwhile, a pruning algorithm is designed to remove inappropriate selections of noise covariance matrices and enhance the performance. The simulation and field test results indicate that the positioning accuracy of the SARSA EKF is better than the traditional EKF and the Q-learning EKF (QLEKF). The positioning accuracy's mean error of the SARSA EKF decreases by 34.32% and 25.95% compared with the traditional EKF and the QLEKF, respectively. And the positioning accuracy's standard deviation of the SARSA EKF decreases by 41.74% and 32.99% compared with the traditional EKF and the QLEKF, respectively.

본 논문은 비가우시안 노이즈가 존재하는 수중환경에서 비선형 필터 기법에 따른 2차원 위치 추정에 관한 연구 내용이다. 최근 위치 추정을 위한 필터로 확장형 칼만필터(EKF: Extended Kalman filter)가 많이 사용되고 있다. 하지만, 수중과 같은 비가우시안 노이즈가 존재하는 비선형 시스템에서는 많은 문제점을 가지고 있다. 따라서 본 논문에서는 상태변이의 예측을 기반으로한 EKF를 대신하여 통계적 발생인자 에 기반을 둔 분포 재해석 기법을 이용한 2차원 파티클필터 (TDPF: Two-Dimension Particle Filter)를 제안한다. 모의 실험을 통하여 Non-Gaussian Noise 가 존재하는 수중환경에서 제안하는 TDPF의 성능을 EKF와 비교분석하였으며 TDPF가 EKF보다 정확한 위치 추정결과를 제공하는 것을 확인하였다. This paper has considered the location estimation problem in two dimension space by using a non-linear filter under non-Gaussian noise in underwater acoustic sensor networks(UASNs). Recently, the extended Kalman filter (EKF) is widely used in location estimation. However, the EKF has a lot of problems in the non-linear system under the non-gaussian noise environment like underwater environment. In this paper, we propose the improved Two-Dimension Particle Filter (TDPF) using the re-interpretation distribution techniques based on the maximum likelihood (ML). Through the simulation, we compared and analyzed the proposed TDPF with the EKF under the non-Gaussian underwater sensor networks. Finally, we determined that the TDPF's result shows more accurate localization than EKF's result.

Autonomous driving is playing more and more an important role, not only in automobile but also in maritime applications. Accurate and robust localization of the local system and the participants nearby is one of the key aspects in applying autonomous surface vehicles. The presented publication is part of the developments of the joint project GALILEOnautic 2, which develops a system for autonomous, cooperative maneuvering of networked vessels in harbors and safety critical areas. In this context, the paper on hand proposes a novel cooperative navigation strategy of a connected autonomous vessels system, which ensures an accurate and robust localization of the networked participants and detects the marine environment as basis of autonomous driving in the maritime field. Using environmental sensors, such as light detecting and ranging (LIDAR) and radio detection and ranging (radar), the objects nearby can be detected and clustered. These detections are then associated with the participants information accessed via network communication. Using the methods, such as similarity comparison and modified Dempster-Schaefer theory, the detections from different resources can be aligned with each other, and the environmental detections can then be classified in different categories. Subsequently, the information from networked participants who possess an accurate geodetic position can be used to enhance the robustness and accuracy of the own local positioning by combining relative distance and relative angle between the own vehicle and a networked participant with a known geodetic position. At this point, an extended Kalman filter is implemented: the information of the neighbor participants, e.g. distance and angle to other objects along with local GNSS information, and velocity measurements, are used to update the EKF states. Before a future maritime test, the proposal is validated through a smaller scenario with three experimental vehicles. A possible networked scenario with networked participants and obstacles is conducted. The results showed, that the proposed implementation can successfully detect, identify, and categorize the objects in the environment. Aided with the cooperative information, the navigation filter can achieve a robust positioning with an accuracy of under one meter, even in the case of an unavailability of GNSS measurements.

This letter presents explicit sub-optimal track-to-track fusion algorithms for Multi-Sensor Systems (MSS) estimating nonlinear processes. The individual tracks in an MSS are correlated due to the presence of a common process noise in the track estimation errors. Herein, we propose recursive formulae for consistent correlation estimation in mildly and highly nonlinear systems that respectively use Extended Kalman Filters (EKF) and Unscented Kalman Filters (UKF) for track estimation. In a mildly nonlinear system, the linearized model employed in the EKF-based MSS architecture offers a correlation propagation formula whose coupling with the optimal track fusion rule generates a sub-optimal fused estimate. On the other hand in highly nonlinear systems, the UKF-based architectures are proven effective for track estimation. The UKF works based on the unscented transform of deterministic sigma points, which is equivalent to the Statistical Linearization Regression (SLR) process. For UKF-based MSS architectures, we propose a consistent correlation propagation recursion according to the SLR technique that will be coupled with the optimal track fusion rule to generate a sub-optimal fused estimate. The performance of the developed fusion algorithms is demonstrated through conducting a statistical test and an average root mean square error analysis.