UAV Platform Research Articles

Extinguishing Real Fires by Fully Autonomous Multirotor UAVs in the MBZIRC 2020 Competition

In this paper, we describe a system for combating real fires with sprayed liquid extinguishing agent using a team of multirotor UAVs. The system design relies on onboard sensors and operates without the need for human intervention throughout its complex mission, from its takeoff to landing. The core UAV platform can estimate its state, self-localize, navigate and locate and extinguishing fires. Thermal and RGB cameras are used, each with a specialized computer vision subsystem and are combined with planar LIDAR for 3D localization of fires on multistory building facades. The system conducts aerial firefighting with a software stack that addresses flight dynamics and sensor limitations and a liquid-spraying subsystem incorporating a rigidly-attached water nozzle. The approach presented in this paper was motivated by the Mohamed Bin Zayed International Robotics Challenge (MBZIRC 2020) firefighting scenario, which focused on coordinated multi-UAV teams that can autonomously combat high-rise building fires. The MBZIRC series places particular emphasis on fast and reliable deployment of robots in realistic conditions. These contests promote development of real-world applications that are greatly needed by society, but which still exceed State-of-the-Art in the robotics community. To our knowledge, our system was the only MBZIRC 2020 contender to extinguish a facade fire successfully in autonomous mode without using an RTK-GNSS system. Our approach contributed to victory in the overall competition and we have now adapted it into an industrial prototype for a firefighting UAV. A video attachment to this paper is available at http://mrs.felk.cvut.cz/fr2020firechallenge-facadefires.

Machine Learning-Based Approaches for Predicting SPAD Values of Maize Using Multi-Spectral Images

Precisely monitoring the growth condition and nutritional status of maize is crucial for optimizing agronomic management and improving agricultural production. Multi-spectral sensors are widely applied in ecological and agricultural domains. However, the images collected under varying weather conditions on multiple days show a lack of data consistency. In this study, the Mini MCA 6 Camera from UAV platform was used to collect images covering different growth stages of maize. The empirical line calibration method was applied to establish generic equations for radiometric calibration. The coefficient of determination (R2) of the reflectance from calibrated images and ASD Handheld-2 ranged from 0.964 to 0.988 (calibration), and from 0.874 to 0.927 (validation), respectively. Similarly, the root mean square errors (RMSE) were 0.110, 0.089, and 0.102% for validation using data of 5 August, 21 September, and both days in 2019, respectively. The soil and plant analyzer development (SPAD) values were measured and applied to build the linear regression relationships with spectral and textural indices of different growth stages. The Stepwise regression model (SRM) was applied to identify the optimal combination of spectral and textural indices for estimating SPAD values. The support vector machine (SVM) and random forest (RF) models were independently applied for estimating SPAD values based on the optimal combinations. SVM performed better than RF in estimating SPAD values with R2 (0.81) and RMSE (0.14), respectively. This study contributed to the retrieval of SPAD values based on both spectral and textural indices extracted from multi-spectral images using machine learning methods.

The Prototype Monitoring System for Pollution Sensing and Online Visualization with the Use of a UAV and a WebRTC-Based Platform.

Nowadays, we observe a great interest in air pollution, including exhaust fumes. This interest is manifested in both the development of technologies enabling the limiting of the emission of harmful gases and the development of measures to detect excessive emissions. The latter includes IoT systems, the spread of which has become possible thanks to the use of low-cost sensors. This paper presents the development and field testing of a prototype pollution monitoring system, allowing for both online and off-line analyses of environmental parameters. The system was built on a UAV and WebRTC-based platform, which was the subject of our previous paper. The platform was retrofitted with a set of low-cost environmental sensors, including a gas sensor able to measure the concentration of exhaust fumes. Data coming from sensors, video metadata captured from 4K camera, and spatiotemporal metadata are put in one situational context, which is transmitted to the ground. Data and metadata are received by the ground station, processed (if needed), and visualized on a dashboard retrieving situational context. Field studies carried out in a parking lot show that our system provides the monitoring operator with sufficient situational awareness to easily detect exhaust emissions online, and delivers enough information to enable easy detection during offline analyses as well.

Urban agriculture as a component of the concept of energy efficient community

Issues of quality and diversity of available food, environmental safety and guarantees of sustainable energy supply are becoming critical for the population of urban areas. Part of the urbanized areas cannot be used directly for food production for environmental reasons, but it is quite suitable for the production of energy from biomass. In order to create strategies for processing organic biomass from urban plantations of parks and squares, it is necessary to determine the tools for obtaining and interpreting data on the available and prospective amount of biomass in cities, which was the purpose of our work. Unmanned aerial vehicle studies were conducted at the Botanical Garden of the National University of Life and Environmental Sciences of Ukraine using the Slantrange 3P multispectral sensor system from an industrialtype UAV platform. The study of squares and lawns of the city was conducted using an archive of satellite images in the visible range of high resolution — 0.5 m/pixel and a specialized agricultural service EO Browser with a resolution of images of 10 m/pixel. Monitoring of the park using the Slantrange 3P complex, with specialized SlantView software, allowed to identify tree crowns and grass cover, while it was not possible to distinguish vines. The best identification results for the Slantrange complex were obtained for the red and infrared measurement channels. Using satellite monitoring data, the possibility of identifying the biomass of trees and shrubs occurs in drought conditions, when the grass on the lawns suffers more than trees and shrubs, apparently due to the developed root system. It is noted that satellite imagery is usually carried out at different angles and, accordingly, fixed both roofs and partially walls of buildings, and accordingly, such imagery is more suitable for assessing the biomass of promising vertical landscaping with vines.

Tree's detection & health's assessment from ultra-high resolution UAV imagery and deep learning

Detection and classification of crown trees properties and assessment of their health status have raised much interest for the scientists from the forest and environmental sciences due to their essential role in landscape ecology and forestry management. The present study proposes a method based on consumer-based UAVs and deep learning techniques for detecting individual orchard trees and assessing key properties characterising their health status. In the proposed scheme, the Mask R-CNN model is used for detecting and mapping each individual tree morphometrical property such as the height and the crown width. Tree's health assessment is based on the use of vegetation indices such as the Visual Atmospheric Resistance Index (VARI) and Green Leaf Index (GLI), computed from the visible spectrum camera mounted on the UAV platform. The use of the proposed approach is demonstrated at five different orchard tree species, namely plum, apricot, walnut, olive, and almond, located in Romania and Greece, computing a series of statistical metrics. Results returned outstanding ability of the algorithm's performance to map the individual trees and assess their health for four out of the five tree species (plum, walnut, apricot, almond) and satisfactory results for the fifth (olive trees). Overall, the study findings highlighted the promising potential of the proposed methodological framework and its scalable potential for wider applicability as a low-cost, effective solution in mapping individual trees properties and health conditions in the field.

UAV Object Detection Model Based on Improved Ghost Module

UAVs with computer vision functions have been widely used in various fields. However, due to the limited memory and computing power of embedded devices, the real-time scene analysis of detection algorithm running on UAV platform is still a challenge. To face the challenge, in this paper, we based on the YOLOV3 detection model by changing the convolution method and using fewer parameters to generate more feature maps. We evaluate our detector on the VisDrone2020 dataset. Compared with the original YOLOV3, we achieved impressive results, including a reduction of 76.38% of the parameter size, 77.95% of floating-point operations, and a considerable accuracy rate and faster inference speed.

Effects of image spatial resolution and statistical scale on water stress estimation performance of MGDEXG: A new crop water stress indicator derived from RGB images

To further explore the performance of the newly proposed crop water stress indicator - the Mean Value of Gaussian Distribution of Excess Green index (MGDEXG) for maize canopy in RGB imagery, studies were conducted in two maize fields with different irrigation levels during the 2015 and 2019 growing seasons. Specifically, the effects of spatial resolution of RGB images collected by both ground and UAV platforms, and segmentation scale of UAV RGB orthophoto on the performance of MGDEXG were investigated, and MGDEXG maps were derived based on UAV RGB orthophoto to monitor maize water status and its inter-field variability. The results show that when the spatial resolution of ground RGB images (2.4 mm) was resized by bilinear interpolation algorithm to 4.8, 9.6, 19.2, 38.4, and 76.8 mm, similar water estimation performances of MGDEXG were observed when compared to the crop water stress index (CWSI), with R2 values ranging 0.80–0.83. However, the processing time per RGB image with 2.4 mm spatial resolution was greatly reduced from 232.26 s to 0.32 s when the resolution was reduced by 32 times, providing a better opportunity to obtain MGDEXG in real-time. When UAV RGB images with two spatial resolutions of 2.7 mm and 14.7 mm were adopted, a poor water stress estimation performance was observed for the lower resolution with R2 of 0.62 and RMSE of 0.12 mol·m-2·s-1 for the maize stomatal conductance. The possible reason could be the errors introduced during the mosaicking process. When segmentation scales of 2 m x 2 m, 4 m x 4 m, 6 m x 6 m, 8 m x 8 m, 10 m x 10 m, and 12 m x 12 m were adopted to crop UAV RGB orthophoto, similar results were also observed. Finally, MGDEXG maps were derived from UAV RGB orthophoto. Overall, this study demonstrated that the water stress estimation performance of MGDEXG index was not affected by image spatial resolution and statistical scale, and MGDEXG maps could be successfully acquired by a UAV RGB remote sensing platform with the advantages of low cost and easy to be adopted by users.

Integrative Geophysical Studies at the Novaya Kurya-1 Cemetery in the Kulunda Steppe

We outline the results of prospection studies at the Novaya Kurya-1 cemetery in the south of Western Siberia, using remote sensing methods such as aerial photography, ground-based magnetometry, high-precision aeromagnetic survey, electromagnetic profi ling, and electrotomography. Original techniques were used to construct relative relief maps, and an inversion of data from ground-based magnetic survey at various altitudes was carried out. The fi rst technique reduces the effect of natural relief, and highlights anthropogenic altitudinal anomalies, making the analysis of digital elevation models more effi cient. The second technique is helpful for assessing the thickness and depth of anomalous magnetic bodies or horizons, not only providing planigraphic information but enabling us to evaluate two- and threedimensional geometric properties of the detected objects. As a result of the analyses, at least 14 kurgans were identifi ed at the cemetery, six of which lack salient outward features. Structural details suggest that most of them date to the Early Scythian time (800–400 BC). On the basis of the interpretation of the results of highly effi cient prospection analyses using the UAV platform, offering the possibility of surveying a large area (about 25 ha), the boundaries of the site were determined. Several features were detected. To identify these, further studies are needed.

Dangerous Area of Tethered UAV Due to Impact

Tethered UAV is composed of a UAV platform, tethered cable and winch. The UAV platform and tethered cable are an interactive whole, and the movement of the UAV platform will change the configuration and tension of the tethered cable. Under extreme external conditions (such as strong airflow) or unprofessional operation, the UAV platform will be prone to extreme movement, which will impact the tethered cable, and in serious cases, the tethered UAV will not work normally. To study the influence on the tethered UAV due to impact, equations for the relation between impact stress and impact parameters are derived and numerically solved to determine the dangerous area of the tethered UAV based on two failure criteria. The obtained results reveal that a smaller elastic modulus and cross-sectional area will lead to less failure of the tethered UAV, and the breaking strength of the cable is vital to the tethered UAV when it is impacted. Engineering suggestions are also provided to make tethered UAVs safer.

Application of UAV technology in environmental impact assessment of mine reclamation project

It is of great significance to carry out environmental impact assessment on the renovated mining area for the assessment of reclamation benefits. The maturing of UAV technology brings new possibilities for environmental assessment. Aiming at the problem of insufficient environmental impact assessment system of mining area renovation project, this paper constructs the environmental impact assessment system of mining area renovation project, and puts forward a set of scheme to carry out the environmental impact assessment of mining area after reclamation based on the data obtained by UAV platform. The research results have certain value for the benefit assessment of comprehensive renovation and ecological reclamation of mining area

Path planning of UAV base station based on deep reinforcement learning

UAV base station platform has become the current research hotspot of assisting ground base station for wireless coverage.At present, the most important issue is how to make path planning to provide the stable communication guarantee for multiple mobile users. In this article, we model the air-to-ground channel to describe the path loss between the UAV platform and the user and build a simulation environment for training based on the OpenAI-GYM architecture. In addition, this paper proposes a reinforcement learning algorithm based on intrinsic rewards, which uses the mean square error of the state prediction results to quantify the novelty of the state. Algorithms enable agents to efficiently carry out strategy iterations. Experiments results showed that our algorithm has a higher score and takes less time.

Classification of Crop Residue Cover in High-Resolution RGB Images Using Machine Learning

HighlightsA machine learning framework estimated residue cover in RGB images taken at three resolutions from 88 locations.The best results primarily used texture features, the RFE-SVM feature selection method, and the SVM classifier.Accounting for shadows and plants plus modifying and optimizing the texture features may improve performance.An automated system developed using machine learning is a viable strategy to estimate residue cover from RGB images obtained with handheld or UAV platforms.Abstract. Maintaining plant residue on the soil surface contributes to sustainable cultivation of arable land. Applying machine learning methods to RGB images of residue could overcome the subjectivity of manual methods. The objectives of this study were to use supervised machine learning while identifying the best feature selection method, the best classifier, and the most effective image feature types for classifying residue levels in RGB imagery. Imagery was collected from 88 locations in 40 row-crop fields in five Missouri counties between early May and late June in 2018 and 2019 using a tripod-mounted camera (0.014 cm pixel-1 ground sampling distance, GSD) and an unmanned aerial vehicle (UAV, 0.05 and 0.14 GSD). At each field location, 50 contiguous 0.3 × 0.2 m region of interest (ROI) images were extracted from the imagery, resulting in a dataset of 4,400 ROI images at each GSD. Residue percentages for ground truth were estimated using a bullseye grid method (n = 100 points) based on the 0.014 GSD images. Representative color, texture, and shape features were extracted and evaluated using four feature selection methods and two classifiers. Recursive feature elimination using support vector machine (RFE-SVM) was the best feature selection method, and the SVM classifier performed best for classifying the amount of residue as a three-class problem. The best features for this application were associated with texture, with local binary pattern (LBP) features being the most prevalent for all three GSDs. Shape features were irrelevant. The three residue classes were correctly identified with 88%, 84%, and 81% 10-fold cross-validation scores for the 2018 training data and 81%, 69%, and 65% accuracy for the 2019 testing data in decreasing resolution order. Converting image-wise data (0.014 GSD) to location residue estimates using a Bayesian model showed good agreement with the location-based ground truth (r2 = 0.90). This initial assessment documents the use of RGB images to match other methods of estimating residue, with potential to replace or be used as a quality control for line-transect assessments. Keywords: Feature selection, Soil erosion, Support vector machine, Texture features, Unmanned aerial vehicle.

Design, development and experimental validation of a lightweight dual-arm aerial manipulator with a COG balancing mechanism

In this work, we present the design, fabrication, and experimental validation of a lightweight, low inertia dual-arm manipulator with a center of gravity (COG) balancing mechanism, specifically designed for aerial manipulation missions. The developed system, consisting of the dual-arm base and two arms with 6 degrees of freedom (DOFs) each, weighs 2.5 kg in total with a maximum payload of 1.0 kg per arm. The dual-arm system is designed such that it can be attached to different multirotors without making major design modifications. In addition, the design of the arms is conceived to decrease the inertia of the arms by utilizing a timing belt-based transmission mechanism. Moreover, the proposed dual-arm design employs prismatic joints to introduce the following distinctive features: 1) ability of each arm to dynamically adjust its COG for better flight performance; 2) fully independent control of each arm for performing different tasks simultaneously; 3) extended workspace and reach of the arms for enhancing operational capability and improving safety during aerial manipulation missions. Since the proposed design has low inertia and a compensation mechanism to deal with the disturbances caused by the COG change, the developed dual-arm system can be mounted on multirotors equipped with standard autopilots, opening the door for the widespread use of dual-arm manipulators with the commercially available UAV platforms. To ensure the robustness of the mechanical structure, an analysis using finite element methods (FEM) is conducted. Extensive experimental flight tests are performed to evaluate the proposed dual-arm design with a hexarotor equipped with a common off-the-shelf autopilot. The experimental results show that the influence of the arms motion over the hexarotor stability is minor in contactless flight due to the low weight and inertia of the proposed dual-arm design. Furthermore, better hovering performance is achieved by exploiting the ability of the proposed design to compensate the dual-arm COG displacement.

Metric contrast of thermal 3D models of large industrial facilities obtained by means of low-cost infrared sensors in UAV platforms

ABSTRACT Monitoring for maintenance or studies of energy efficiency in buildings, large infrastructure, industrial facilities, etc., are common nowadays. These kind of studies are developed with inspections which determine the state of the facilities that are analysed. The difficulty is increased along with the size and complexity of the facility itself, and even more when the attribute to be surveyed is not noticeable or responsive for the human eye. In recent years, a series of techniques that rely on different sensors mounted on UAVs allow detecting problems that are associated with facilities of large dimensions. Almost all of them work in the visible band (RGB), but the generation of thermal 3D models permits detecting any heat anomaly related to the functioning of these facilities. This research proposes a methodology and workflow for the generation and Metric Contrast of Thermal Models (MCTM). This methodology is metrically applied to a mining-industrial facility in which thermal conditions have great influence for a proper functioning. For this metric contrast, several distances have been measured in the field and compared to those obtained from the models. The average difference between the true magnitude and those obtained from the RGB and thermal models are 5 and 31 cm, and their standard deviations are 7 and 29 cm, respectively. The comparison between the RGB and the thermal model provides an average distance between points is 0.19 m, and for 75% of the points the distance is lesser than 0.35 m. Although the RGB model is more accurate, the precision of the thermal model is enough for the objectives set.

Wireless Charging and Endurance Platform of Patrol UAV Based on Inductive Power Collection of Transmission Line

This project designs a wireless charging and endurance platform for patrol UAV based on inductive power collection. The main function is to solve the problems of short dead time and continuous inspection of ordinary patrol UAV. The system is mainly composed of transmission line inductive energy taking coil, inductive power supply, energy storage battery module, wireless charging module, energy management module and patrol UAV with wireless charging function.

Eden Library: A long-term database for storing agricultural multi-sensor datasets from UAV and proximal platforms

In modern agriculture, visual recognition systems based on deep learning are arising to allow autonomous machines to execute field operations in crops. However, for obtaining high performances, these methods need high amounts of data, which are usually scarce in agriculture. The main reason is that building an agricultural dataset covering exhaustively a specific problem is challenging, as visual characteristics of the symptoms may change, and there is a high dependency on environmental factors, such as temperature, humidity and light conditions. Therefore, an efficient methodology is necessary to consistently cover the entire workflow for creating an agricultural dataset, from the image acquisition to its online publication. This paper presents the Eden Library, a platform for contributing to this existing gap in open access crop/plant databases covering proximal and aerial images. The complete workflow on the design and deployment of the platform is also explained and discussed. This workflow is relevant because the provided datasets are thought to be maintained and enriched along the time, and they do not just remain as a static research output covering only specific species, growth stages, and conditions. The image annotations of plants and symptoms are provided, saving users from manually annotating images. Currently, the Eden Library covers 15 different crops, 9 weeds and 30 disorders (pests, diseases and nutrient deficiencies). Eden Library aspires to close this gap by providing a large and diversified image collection of plants, organized in a consistent manner, in order to boost further vision-based and AI-enabled field applications.

Remote sensing of agrophytocenoses from the UAV platform to assess the level of plant nutrition

Unmanned (remotely controlled) aircrafts (UAVs) are innovative equipment for monitoring fields that are free from a lot of the disadvantages of satellites such as availability, low cost, and high image resolution. However, the quality, reproducibility and suitability of spectral monitoring data of plantations for crop management processes remain topical issues. Since spectral monitoring is a necessary component in the concept of crop management, the development of a methodology for assessing the suitability of remote monitoring spectral data for the calculation of agrochemical practices was the purpose of the work. According to the publications, the dependence of the number of pixels on the values of the intensity of color components for plants and soil is described by the Gaussian distribution. Deviation from such distribution is caused by the imposing of distributions from various objects fixed on a photo. The experimental test was carried out on the basis of wheat, using the results obtained during 2017–2020 when considering the stresses of nutrient deficiency. UAVs with a specialized Slantrange spectral complex with standard Slantview software and an FC200 visible spectrum camera (from the Phantom 2 UAV) were used for monitoring. The investigation found experimental evidence that the pixel distribution of plantations on the example of the wheat crop is described by the Gaussian distribution. It was found that the analysis of the correspondence of the nature of the distribution on the spectral channels, namely the presence of several max peaks that affects the value of the distribution maximum may indicate the presence of foreign inclusions or a transitional stage of vegetation. The suitability of the data can be assessed on the basis of the reference values of the width of the distribution on the spectral channels. This determines the feasibility of introducing in the sets of regular vegetation indices of geographic information systems additional packages that reflect the spectral channels.

Weather Sensing in an Urban Environment with the Use of a UAV and WebRTC-Based Platform: A Pilot Study.

Thanks to IoT, Internet access, and low-cost sensors, it has become possible to increase the number of weather measuring points; hence, the density of the deployment of sources that provide weather data for the needs of large recipients, for example, weather web services or smart city management systems, has also increased. This paper presents a flying weather station that carries out measurements of two weather factors that are typically included in weather stations (ambient temperature and relative humidity), an often included weather factor (atmospheric pressure), and a rarely included one (ultraviolet index). In our solution, the measurements are supplemented with a visual observation of present weather phenomena. The flying weather station is built on a UAV and WebRTC-based universal platform proposed in our previous paper. The complete, fully operational flying weather station was evaluated in field studies. Experiments were conducted during a 6-month period on days having noticeably different weather conditions. Results show that weather data coming from the flying weather station were equal (with a good approximation) to weather data obtained from the reference weather station. When compared to the weather stations described in the literature (both stationary weather stations and mobile ones), the proposed solution achieved better accuracy than the other weather stations based on low-cost sensors.

How Can Unmanned Aerial Vehicles Be Used for Detecting Weeds in Agricultural Fields?

Weeds are among the most harmful abiotic factors in agriculture, triggering significant yield loss worldwide. Remote sensing can detect and map the presence of weeds in various spectral, spatial, and temporal resolutions. This review aims to show the current and future trends of UAV applications in weed detection in the crop field. This study systematically searched the original articles published from 1 January 2016 to 18 June 2021 in the databases of Scopus, ScienceDirect, Commonwealth Agricultural Bureaux (CAB) Direct, and Web of Science (WoS) using Boolean string: “weed” AND “Unmanned Aerial Vehicle” OR “UAV” OR “drone”. Out of the papers identified, 144 eligible studies did meet our inclusion criteria and were evaluated. Most of the studies (i.e., 27.42%) on weed detection were carried out during the seedling stage of the growing cycle for the crop. Most of the weed images were captured using red, green, and blue (RGB) camera, i.e., 48.28% and main classification algorithm was machine learning techniques, i.e., 47.90%. This review initially highlighted articles from the literature that includes the crops’ typical phenology stage, reference data, type of sensor/camera, classification methods, and current UAV applications in detecting and mapping weed for different types of crop. This study then provides an overview of the advantages and disadvantages of each sensor and algorithm and tries to identify research gaps by providing a brief outlook at the potential areas of research concerning the benefit of this technology in agricultural industries. Integrated weed management, coupled with UAV application improves weed monitoring in a more efficient and environmentally-friendly way. Overall, this review demonstrates the scientific information required to achieve sustainable weed management, so as to implement UAV platform in the real agricultural contexts.

Verification of the Geometrical Representation of Buildings in Cadastre Using UAV Photogrammetry

The aim of this study was to determine the potential benefits of using UAV-based data to verify the accuracy of the geometry of the outlines of buildings registered in a cadastral database, and to investigate the potential use of data from UAV platforms in measuring the buildings.The authors examined whether the acquisition of data from UAVs can replace field measurements or only support them. The advantages and disadvantages of such data acquisition methods, compared with more well-established measurement methods, were identified.