FIGURE 6 - available via license: Creative Commons Attribution 4.0 International
Content may be subject to copyright.
Source publication
Mapping, as the back-end of perception and the front-end of path planning in the modern UAV navigation system, draws our interest. Considering the requirements of UAV navigation and the features of the current embedded computation platforms, we designed and implemented a novel multilayer mapping framework. In this framework, we divided the map into...
Context in source publication
Context 1
... occupied/non-occupied measurements information is then used to update the occupancy probability (represented using log-odds value) on the local map (Equation 8 and Table 1). The local-global map thread contains two independent modules to generate the project 2D map and local ESDFs ( Figure 6). The projected 2D map projects all of the occupied voxels to the ground level. ...
Citations
... Another way to address the problem of large environments is efficiently mapping the environment for the computation-friendly navigation of UAVs. One such effort was presented by [64], where a mapping framework comprised three layers: awareness, local, and global. This study is also offered as an open source for the research community. ...
... ROS GAZEBO [9,17,19,21,26,30,54,56,64,74,83] MATLAB/Simulink [13,14,30,33,50,53,57,58,67,75,79,80,85,87] Python (2.×, 3.×, PyCharm etc.) [55] V-REP [17,24,26] Kestrel (ViDAR) [50] Air-Learning [9] AirSim (Unreal Engine) [9,52] Flight Gear Simulator [58] QGroundControl [30] ArduPilot [32,62] PIXHAWK [21,30,62,64,83] HK Pilot 32 [85] RaspberryPie [35] ODroidXU [43] Beaglebone [62] Drones 2023, 7, 118 ...
... ROS GAZEBO [9,17,19,21,26,30,54,56,64,74,83] MATLAB/Simulink [13,14,30,33,50,53,57,58,67,75,79,80,85,87] Python (2.×, 3.×, PyCharm etc.) [55] V-REP [17,24,26] Kestrel (ViDAR) [50] Air-Learning [9] AirSim (Unreal Engine) [9,52] Flight Gear Simulator [58] QGroundControl [30] ArduPilot [32,62] PIXHAWK [21,30,62,64,83] HK Pilot 32 [85] RaspberryPie [35] ODroidXU [43] Beaglebone [62] Drones 2023, 7, 118 ...
UAVs have been contributing substantially to multi-disciplinary research and around 70% of the articles have been published in just about the last five years, with an exponential increase. Primarily, while exploring the literature from the scientific databases for various aspects within the autonomous UAV path planning, such as type and configuration of UAVs, the complexity of their environments or workspaces, choices of path generating algorithms, nature of solutions and efficacy of the generated paths, necessitates an increased number of search keywords as a prerequisite. However, the addition of more and more keywords might as well curtail some conducive and worthwhile search results in the same pursuit. This article presents a Systematic Literature Review (SLR) for 20 useful parameters, organized into six distinct categories that researchers and industry practitioners usually consider. In this work, Web of Science (WOS) was selected to search the primary studies based on three keywords: “Autonomous” + “Path Planning” + “UAV” and following the exclusion and inclusion criteria defined within the SLR methodology, 90 primary studies were considered. Through literature synthesis, a unique perspective to see through the literature is established in terms of characteristic research sectors for UAVs. Moreover, open research challenges from recent studies and state-of-the-art contributions to address them were highlighted. It was also discovered that the autonomy of UAVs and the extent of their mission complexities go hand-in-hand, and the benchmark to define a fully autonomous UAV is an arbitral goal yet to be achieved. To further this quest, the study cites two key models to measure a drone’s autonomy and offers a novel complexity matrix to measure the extent of a drone’s autonomy. Additionally, since preliminary-level researchers often look for technical means to assess their ideas, the technologies used in academic research are also tabulated with references.
... The maps were projected in 2D and local Euclidean signed distance fields (ESDFs) [81]. The distance between the ESDFs represents the Euclidean distance to the nearest occupied voxel [82]. The path-planning Fuxi kit consists of two parallel running planners: global and local planners. ...
In recent years, unmanned aerial vehicles (UAVs), commonly known as drones, have gained increasing interest in both academia and industries. The evolution of UAV technologies, such as artificial intelligence, component miniaturization, and computer vision, has decreased their cost and increased availability for diverse applications and services. Remarkably, the integration of computer vision with UAVs provides cutting-edge technology for visual navigation, localization, and obstacle avoidance, making them capable of autonomous operations. However, their limited capacity for autonomous navigation makes them unsuitable for global positioning system (GPS)-blind environments. Recently, vision-based approaches that use cheaper and more flexible visual sensors have shown considerable advantages in UAV navigation owing to the rapid development of computer vision. Visual localization and mapping, obstacle avoidance, and path planning are essential components of visual navigation. The goal of this study was to provide a comprehensive review of vision-based UAV navigation techniques. Existing techniques have been categorized and extensively reviewed with regard to their capabilities and characteristics. Then, they are qualitatively compared in terms of various aspects. We have also discussed open issues and research challenges in the design and implementation of vision-based navigation techniques for UAVs.
... Data from two or more sensors about the same target could be fused to enhance confidence in the data. For example, the data pertaining to overlapping landmarks in visual sensors are considered to be redundant while the data on the same landmark captured by two sensors with different fields of view are considered cooperative [92]. (vii) Reliability issues: Often, the sensor data are not just uncertain, but could also be unreliable. ...
... In recursive estimations, drift occurs when errors are accumulated over time. The error in prior estimations has a significant impact on the new estimations [92]. EKF is effective when computing resources are limited or the state dimension is relatively small [139]. ...
This article presents a survey of simultaneous localization and mapping (SLAM) and data fusion techniques for object detection and environmental scene perception in unmanned aerial vehicles (UAVs). We critically evaluate some current SLAM implementations in robotics and autonomous vehicles and their applicability and scalability to UAVs. SLAM is envisioned as a potential technique for object detection and scene perception to enable UAV navigation through continuous state estimation. In this article, we bridge the gap between SLAM and data fusion in UAVs while also comprehensively surveying related object detection techniques such as visual odometry and aerial photogrammetry. We begin with an introduction to applications where UAV localization is necessary, followed by an analysis of multimodal sensor data fusion to fuse the information gathered from different sensors mounted on UAVs. We then discuss SLAM techniques such as Kalman filters and extended Kalman filters to address scene perception, mapping, and localization in UAVs. The findings are summarized to correlate prevalent and futuristic SLAM and data fusion for UAV navigation, and some avenues for further research are discussed.
The Dynamic Window Approach (DWA) is a popular method for Unmanned Aerial Vehicle (UAV) navigation and localization in unknown environments. It combines Dynamic Programming (DP) with a Probabilistic Route Mapping (PRM) algorithm to provide efficient path planning and obstacle avoidance. DWA can handle a wide range of obstacles, including dynamic and uncertain ones, making it highly reliable. The approach utilizes dynamic programming to compute the optimal path based on the UAV's current state and the known environment. It also employs a hybrid probabilistic route mapping algorithm to estimate the location and movement of unknown obstacles. By combining these techniques, DWA enables the UAV to navigate through complex environments efficiently. One of DWA's key strengths is its ability to handle non-holonomic constraints, such as the limited turning radius of a mobile UAV. It achieves this by defining a dynamic window that determines the feasible set of motions for the UAV at any given time and adjusts the path accordingly. Compared to other popular methods like the Rapidly Exploring Random Trees (RRT) algorithm, DWA outperforms in terms of path planning and obstacle avoidance. It overcomes the limitations imposed by the size of autonomous mobile UAVs by considering the relationship between the robot's dimensions and obstacles in the open space. To enhance sensing and prediction of the surroundings, a laser range finder is utilized in DWA, particularly to handle curved structures or box-canyon formations. This, along with the Dynamic Programming (DP) algorithm, optimizes the path by considering the gathered information. The proposed approach addresses the local minima problem through a strategy to identify the effective path region. Theoretical studies and simulations demonstrate the efficiency and superiority of DWA. In summary, the Dynamic Window Approach is an efficient method for UAV navigation and localization in unknown environments. By combining dynamic programming, probabilistic route mapping, and considering non-holonomic constraints, it provides reliable path planning and obstacle avoidance. Its ability to handle various obstacles, including dynamic ones, sets it apart from other methods, making it highly valuable for UAV applications.
In this article, we present a stereo visual inertial pose estimation method based on feedforward and feedbacks. Compared to the widely used filter-based or optimization-based approaches, the proposed method only stores the most recent pose and measurements and thus can achieve fast processing. A gradient decreased feedback, a roll-pitch feedforward, and a bias estimation feedback, are introduced to fuse the vision and the inertial measurements. This system, which is called feedforward and feedback based visual inertial system (FVIS), is evaluated on the popular European robotics challenge micro aerial vehicle (EuRoC MAV) dataset. FVIS achieves high accuracy and robustness with respect to existing visual inertial simultaneous localization and mapping (SLAM) approaches. FVIS has also been implemented and tested on a unmanned aerial vehicle (UAV) platform. The source code developed during this study is available publicly.
In this paper, a novel UAV autonomous navigation system is proposed, which uses DNN to extract scene features for position matching and optical flow estimation. This method can correct position drift of IMU under GNSS-denied environment. A realistic simulation test platform is designed to validate the fusion navigation system, and we deploy the system on an airborne embedded GPU hardware. Optimization strategies for hardware are designed to significantly improve operational performance, which make it become more practical in engineering. Field experimental results further verify its effectiveness of proposed system. Compared with traditional methods, it is more robust and has higher positioning accuracy.
KeywordsAutonomous navigationDeep neural networkGNSS-denied environmentUnmanned aerial vehicle
Algorithms for reconstruction of three-dimensional semantic maps are an important element of on-board vehicle computer vision systems. Such maps can be used in simulation environments and to generate the so-called HD maps needed for path planning and vehicle navigation. The paper presents an analysis of modern methods for semantic map reconstruction based on sequences of 3D point clouds, including noisy ones. The Kimera Semantics method, modified approaches VDB Fusion, Puma and ALeGO-LOAM with Interactive SLAM are compared. We have developed a novel approach for quantitatively estimation the quality of 3D maps and successfully applied it using the open dataset SemanticKITTI. It allows us to take into account the features of generated 3D map mesh and semantic labels in order to obtain a more informative metric.
