GPS-denied Scenarios Research Articles

Enhancing vehicle localization by matching HD map with road marking detection

Accurate and robust vehicle localization, a fundamental task for autonomous driving, is still a challenging problem especially in GPS-denied scenarios. Recently, high definition (HD) map suggests a promising solution. However, the matching between online sensed data and HD map is difficult and time-consuming. In our work, road markings are selected as landmarks due to salient appearance features. Based on the detection results of road markings from the vehicle-borne images, the points of edge lines are employed to fit straight lines with RANSAC for outlier removal. The distances between the ego-vehicle and the fitted edge lines can be computed with camera-vehicle calibration in advance. Subsequently, the point-to-line distances from the sensed data are mapped into global linear constraints on the vehicle’s positions with the support of a lane-level HD map, which provides centimeter-level coordinates of road markings. Finally, the distances and the localization from the integrated navigation system (INS) are fused with the proposed linear Kalman filter based on the second-order Markov model (KF-MM2). The proposed method has been verified in different daily driving scenarios. Experimental results demonstrate that our method can achieve good performance with an average localization error of 0.53 m and a standard deviation of 0.17 m.

Vision-Based Autonomous Landing for Unmanned Aerial and Ground Vehicles Cooperative Systems

In this work, we consider a cooperative system in which UAVs perform long-distance missions with assistance of unmanned ground vehicles (UGVs) for battery charging. We propose an autonomous landing scheme for the UAV to land on a mobile UGV with high precision by leveraging multiple-scale Quick Response (QR)-codes for different altitudes. These QR-codes support the acquisition of the relative distance and direction between the UGV and UAV. As a result, the UGV is not required to report its accurate state to the UAV. In contrast to the conventional landing algorithms based on the global positioning system (GPS), the proposed vision-based autonomous landing algorithm is available for both outdoor and GPS-denied scenarios. To cope with the challenge on the moving platform landing, a quadratic programming (QP) problem is formulated to design the velocity controller by combining a control barrier function (CBF) and a control Lyapunov function (CLF). Finally, both extensive computer simulation and a prototype of the proposed system are shown to confirm the feasibility of the proposed system and landing scheme.

An aerial/ground robot team for autonomous firefighting in urban GNSS-denied scenarios

This paper presents a framework for urban firefighting with a heterogeneous aerial/ground robot team. The system was developed to address Challenge 3 of the MBZIRC’20. The challenge required autonomously detecting, locating, and extinguishing fires on multiple interior floors and on exterior facade and ground surfaces. Our multi-robot system consists of a robot team of up to three Unmanned Aerial Vehicles (UAV) and one Unmanned Ground Vehicle (UGV). We describe the main hardware and software components for UAV and UGV platforms and present the system’s main algorithmic components: a 3D LIDAR-based Mapping and Localization module able to work in GPS-denied scenarios, a Global Planner and a fast local re-planning system for robot navigation, Infrared-Based Perception and robot actuation control for fire extinguishing, and Mission Executive and coordination module based on Behavior Trees. Finally we describe our results from the competition, where the system worked fully autonomously and scored in all the trials performed. The presented system ended in 7 th place (out of 20 teams) in Challenge 3 and 5 th place (out of 17 teams) in the Grand Challenge’s Challenge 3 portion.

Mobile Robot 6-D Localization Using 3-D Gaussian Mixture Maps in GPS-Denied Environments

The mobile robot 6D localization process in GPS-denied scenarios, e.g., in a cave or a mine, is a challenging problem. This paper presents the modification of a well-known literature method using Gaussian mixture maps to determine the robot pose in rough terrain and outdoor non-urbanized environment. To improve the accuracy in rough 3D terrain, I extended the terrain description from a 1D Gaussian mixture model (1DGMM) to a 3D Gaussian mixture model (3DGMM) combining terrain height and terrain inclination in two orthogonal directions. Using this approach, one can maintain centimeter accuracy even in the most demanding workspaces. Moreover, I replaced the 3D scan matching process combined with ground-based smoothing by 6D scan processing (therefore, the 3DGMM method is capable for both land vehicles and flying robots) using the point-by-point EKF based on [12]. Thanks to the point-by-point correction approach, it is possible to run the system in real-time and process a full point cloud without GPU-based computing up to 300 000 points per second using small computers dedicated to mobile robots.

Ground and Aerial Collaborative Mapping in Urban Environments

A heterogeneous multi-robot system consisting of Unmanned Ground Vehicles (UGVs) and Unmanned Aerial Vehicles (UAVs) have advantages over a single-robot system in efficiency and flexibility, enabling them to perform a larger range of tasks. To allow heterogeneous platforms to work together in GPS-denied scenarios, it is crucial to build a complete 3D map of the environment. In this letter, a novel method is presented to perform ground and aerial collaborative mapping leveraging visual and range data collected by cameras and 3D LiDAR sensors. In the proposed system, a visual-LiDAR ego-motion estimation module that considers point, line and planar constraints can provide robust odometry information. Thumbnail images representing obstacle outlines are generated and descriptors are extracted using a neural network to help perform data association between separate runs. Map segments and the robot poses are organized together and are updated during a pose graph optimization procedure. The proposed ground-aerial collaborative mapping approach is evaluated on both synthetic and real-world datasets comparing with other methods. Experiment results demonstrate that our method can achieve outstanding mapping results.

PSD – probabilistic algorithm for mobile robot 6D localization without natural and artificial landmarks based on 2.5D map and a new type of laser scanner in GPS-denied scenarios

This paper presents an approach to mobile robot 6D localization based on a 3D laser scanner in GPS-denied scenarios. Commonly, 6D localization using laser scanners is performed with the use of extraction and association of the features or by comparison of the whole scans (very often off-line) using the ICP algorithm or its modifications. However, in some unstructured non-urbanized rough terrain environments, feature extraction does not seem to be reliable enough. For such kind of environment, we present a new method to mobile robot localization in GPS-denied applications, called PSD (Point-to-Surfel Distance). Unlike state of the art localization methods using laser scanners, we consider every single laser scanner measurement as an observation and use Point-to-Surfel Distance for correction of position and orientation of the robot. Mobile robot localization is based on a specific representation of the terrain in the 2.5D surfel map (terrain height and inclination). The simulation tests compared our method using extended Kalman filter (EKF) and single laser scanner measurements with an up-to-date method using particle filter (PF) and comparing the scan lines with the reference map and with another method using Gaussian mixture maps. The tests confirmed that the proposed method provides satisfying results for GPS-denied scenarios in rough terrain without extractable landmarks and our method is thirty times faster than the PF method (serial implementation). KITTI benchmark tests and real terrain experiments confirmed its usefulness and advantages as well.

Autonomous landing of small unmanned aerial rotorcraft based on monocular vision in GPS-denied area

Focusing on the low-precision attitude of a current small unmanned aerial rotorcraft at the landing stage, the present paper proposes a new attitude control method for the GPS-denied scenario based on the monocular vision. Primarily, a robust landmark detection technique is developed which leverages the well-documented merits of supporting vector machines (SVMs) to enable landmark detection. Then an algorithm of nonlinear optimization based on Newton iteration method for the attitude and position of camera is put forward to reduce the projection error and get an optimized solution. By introducing the wavelet analysis into the adaptive Kalman filter, the high frequency noise of vision is filtered out successfully. At last, automatic landing tests are performed to verify the method's feasibility and effectiveness.

UAV helicopter relative state estimation for autonomous landing on moving platforms in a GPS-denied scenario

The objective of this work is to describe an innovative relative position and velocity state estimation approach for rotatory wing UAV linked with a rope. An autonomous landing of a UAV helicopter in a static and in mobile platform in absence of GPS are presented as application of this technology. The relative estimation is obtained using a device that provides rope orientation, and tension information, an altimeter and an IMU. All this information is combined using a sensor fusion strategy.

Pedestrian dead reckoning for MARG navigation using a smartphone

Abstract The demand for navigating pedestrian by using a hand-held mobile device increased remarkably over the past few years, especially in GPS-denied scenario. We propose a new pedestrian dead reckoning (PDR)-based navigation algorithm by using magnetic, angular rate, and gravity (MARG) sensors which are equipped in existing commercial smartphone. Our proposed navigation algorithm consists of step detection, stride length estimation, and heading estimation. To eliminate the gauge step errors of the random bouncing motions, we designed a reliable algorithm for step detection. We developed a BP neural network-based stride length estimation algorithm to apply to different users. In response to the challenge of magnetic disturbance, a quaternion-based extended Kalman filter (EKF) is introduced to determine the user's heading direction for each step. The performance of our proposed pedestrian navigation algorithm is verified by using a smartphone in providing accurate, reliable, and continuous location tracking services.

Self-organised localisation in indoor environments using the ALF framework

We present and discuss the autonomous localisation framework ALF, a self-organising indoor localisation environment. Location awareness is an important property for a growing number of applications. GPS is frequently used to provide this information in outdoor environments, but this is not applicable for indoor applications. There have been many approaches to solve the localisation problem for those GPS-denied scenarios. However, many of them are limited to certain hardware restrictions or do not provide robust self-localisation in dynamic real world application. ALF is a complete and modular framework based on minimal hardware requirements. The system is not only capable to deploying itself autonomously in unknown environments and offering position information among the participants, but it also supports accurate real-time localisation to customers. The concepts allows to remove or to add features e.g., the heading of nodes or certain real-time capabilities as the scenario demands or the even the used hardware changes. The awareness and handling of measurement errors, especially in non-line of sight NLOS cases, is an essential criterion for a real world application.

Prototype personal navigation system

Honeywell Laboratories recently funded the development of a prototype personal navigation system based on MEMS technologies. The system components include a MEMS inertial measurement unit, a three-axis magnetometer, a barometric pressure sensor, and a SAASM GPS receiver. The system also uses Honeywell's human motion-based pedometry algorithm. The navigation process is based on a strap-down inertial navigator aided by feedback from a Kalman filter using typical measurements from the GPS, magnetometer and barometer when available. A key innovation is the addition of an independent measurement of distance traveled based on the use of a human motion algorithm. The navigation system combines the best features of dead reckoning and inertial navigation, resulting in positioning performance exceeding that achieved with either method alone. Subsequent to the Honeywell effort, DARPA funded an individual Personal Inertial Navigation System (iPINS) seedling program. Honeywell worked to improve the baseline personal navigation system with the objective of demonstrating the feasibility of reliably achieving navigation accuracy < 1 % of distance traveled in GPS-denied scenarios. In addition, an analysis was conducted to determine the benefit of incorporating terrain correlation into the personal navigation system. The results of this analysis indicate that overall navigation accuracy can be significantly improved through the application of terrain correlation. This presents an are presented. In addition, conclusions from the terrain correlation analysis conducted under the iPINS seedling program are included.