Assessing the Performance of a Low-Cost Thermal Camera in Proximal and Aerial Conditions

Mobile Robots for In-Process Monitoring of Aircraft Systems Assemblies

Currently, systems installed on large-scale aerospace structures are manually equipped by trained operators. To improve current methods, an automated system that ensures quality control and process adherence could be used. This work presents a mobile robot capable of autonomously inspecting aircraft systems and providing feedback to workers. The mobile robot can follow operators and localise the position of the inspection using a thermal camera and 2D lidars. While moving, a depth camera collects 3D data about the system being installed. The in-process monitoring algorithm uses this information to check if the system has been correctly installed. Finally, based on these measurements, indications are shown on a screen to provide feedback to the workers. The performance of this solution has been validated in a laboratory environment, replicating a trailing edge equipping task. During testing, the tracking and localisation systems have proven to be reliable. The in-process monitoring system was also found to provide accurate feedback to the operators. Overall, the results show that the solution is promising for industrial applications.

Factors Influencing Temperature Measurements from Miniaturized Thermal Infrared (TIR) Cameras: A Laboratory-Based Approach

The workflow for estimating the temperature in agricultural fields from multiple sensors needs to be optimized upon testing each type of sensor’s actual user performance. In this sense, readily available miniaturized UAV-based thermal infrared (TIR) cameras can be combined with proximal sensors in measuring the surface temperature. Before the two types of cameras can be operationally used in the field, laboratory experiments are needed to fully understand their capabilities and all the influencing factors. We present the measurement results of laboratory experiments of UAV-borne WIRIS 2nd GEN and handheld FLIR E8-XT cameras. For these uncooled sensors, it took 30 to 60 min for the measured signal to stabilize and the sensor temperature drifted continuously. The drifting sensor temperature was strongly correlated to the measured signal. Specifically for WIRIS, the automated non-uniformity correction (NUC) contributed to extra uncertainty in measurements. Another problem was the temperature measurement dependency on various ambient environmental parameters. An increase in the measuring distance resulted in the underestimation of surface temperature, though the degree of change may also come from reflected radiation from neighboring objects, water vapor absorption, and the object size in the field of view (FOV). Wind and radiation tests suggested that these factors can contribute to the uncertainty of several Celsius degrees in measured results. Based on these indoor experiment results, we provide a list of suggestions on the potential practices for deriving accurate temperature data from radiometric miniaturized TIR cameras in actual field practices for (agro-)environmental research.

Predicting Soil Water Content on Rainfed Maize through Aerial Thermal Imaging

Restrictions on soil water supply can dramatically reduce crop yields by affecting the growth and development of plants. For this reason, screening tools that can detect crop water stress early have been long investigated, with canopy temperature (CT) being widely used for this purpose. In this study, we investigated the relationship between canopy temperature retrieved from unmanned aerial vehicles (UAV) based thermal imagery with soil and plant attributes, using a rainfed maize field as the area of study. The flight mission was conducted during the late vegetative stage and at solar noon, when a considerable soil water deficit was detected according to the soil water balance model used. While the images were being taken, soil sampling was conducted to determine the soil water content across the field. The sampling results demonstrated the spatial variability of soil water status, with soil volumetric water content (SVWC) presenting 10.4% of variation and values close to the permanent wilting point (PWP), reflecting CT readings that ranged from 32.8 to 40.6 °C among the sampling locations. Although CT correlated well with many of the physical attributes of soil that are related to water dynamics, the simple linear regression between CT and soil water content variables yielded coefficients of determination (R2) = 0.42, indicating that CT alone might not be sufficient to predict soil water status. Nonetheless, when CT was combined with some soil physical attributes in a multiple linear regression, the prediction capacity was significantly increased, achieving an R2 value = 0.88. This result indicates the potential use of CT along with certain soil physical variables to predict crop water status, making it a useful tool for studies exploring the spatial variability of in-season drought stress.

Transformations in the Photogrammetric Co-Processing of Thermal Infrared Images and RGB Images

The photogrammetric processing of thermal infrared (TIR) images deals with several difficulties. TIR images ordinarily have low-resolution and the contrast of the images is very low. These factors strongly complicate the photogrammetric processing, especially when a modern structure from motion method is used. These factors can be avoided by a certain co-processing method of TIR and RGB images. Two of the solutions of co-processing were suggested by the authors and are presented in this article. Each solution requires a different type of transformation–plane transformation and spatial transformation. Both types of transformations are discussed in this paper. On the experiments that were performed, there are presented requirements, advantages, disadvantages, and results of the transformations. Both methods are evaluated mainly in terms of accuracy. The transformations are presented on suggested methods, but they can be easily applied to different kinds of methods of co-processing of TIR and RGB images.

Comparison of Accuracy of Surface Temperature Images from Unmanned Aerial Vehicle and Satellite for Precise Thermal Environment Monitoring of Urban Parks Using In Situ Data

Rapid urbanization has led to several severe environmental problems, including so-called heat island effects, which can be mitigated by creating more urban green spaces. However, the temperature of various surfaces differs and precise measurement and analyses are required to determine the “coolest” of these. Therefore, we evaluated the accuracy of surface temperature data based on thermal infrared (TIR) cameras mounted on unmanned aerial vehicles (UAVs), which have recently been utilized for the spatial analysis of surface temperatures. Accordingly, we investigated land surface temperatures (LSTs) in green spaces, specifically those of different land cover types in an urban park in Korea. We compared and analyzed LST data generated by a thermal infrared (TIR) camera mounted on an unmanned aerial vehicle (UAV) and LST data from the Landsat 8 satellite for seven specific periods. For comparison and evaluation, we measured in situ LSTs using contact thermometers. The UAV TIR LST showed higher accuracy (R2 0.912, root mean square error (RMSE) 3.502 °C) than Landsat TIR LST accuracy (R2 value lower than 0.3 and RMSE of 7.246 °C) in all periods. The Landsat TIR LST did not show distinct LST characteristics by period and land cover type; however, grassland, the largest land cover type in the study area, showed the highest accuracy. With regard to the accuracy of the UAV TIR LST by season, the accuracy was higher in summer and spring (R2 0.868–0.915, RMSE 2.523–3.499 °C) than in autumn and winter (R2 0.766–0.79, RMSE 3.834–5.398 °C). Some land cover types (concrete bike path, wooden deck) were overestimated, showing relatively high total RMSEs of 4.439 °C and 3.897 °C, respectively, whereas grassland, which has lower LST, was underestimated—showing a total RMSE of 3.316 °C. Our results showed that the UAV TIR LST could be measured with sufficient reliability for each season and land cover type in an urban park with complex land cover types. Accordingly, our results could contribute to decision-making for urban spaces and environmental planning in consideration of the thermal environment.

Mapping the Groundwater Level and Soil Moisture of a Montane Peat Bog Using UAV Monitoring and Machine Learning

One of the best preconditions for the sufficient monitoring of peat bog ecosystems is the collection, processing, and analysis of unique spatial data to understand peat bog dynamics. Over two seasons, we sampled groundwater level (GWL) and soil moisture (SM) ground truth data at two diverse locations at the Rokytka Peat bog within the Sumava Mountains, Czechia. These data served as reference data and were modeled with a suite of potential variables derived from digital surface models (DSMs) and RGB, multispectral, and thermal orthoimages reflecting topomorphometry, vegetation, and surface temperature information generated from drone mapping. We used 34 predictors to feed the random forest (RF) algorithm. The predictor selection, hyperparameter tuning, and performance assessment were performed with the target-oriented leave-location-out (LLO) spatial cross-validation (CV) strategy combined with forward feature selection (FFS) to avoid overfitting and to predict on unknown locations. The spatial CV performance statistics showed low (R2 = 0.12) to high (R2 = 0.78) model predictions. The predictor importance was used for model interpretation, where temperature had strong impact on GWL and SM, and we found significant contributions of other predictors, such as Normalized Difference Vegetation Index (NDVI), Normalized Difference Index (NDI), Enhanced Red-Green-Blue Vegetation Index (ERGBVE), Shape Index (SHP), Green Leaf Index (GLI), Brightness Index (BI), Coloration Index (CI), Redness Index (RI), Primary Colours Hue Index (HI), Overall Hue Index (HUE), SAGA Wetness Index (TWI), Plan Curvature (PlnCurv), Topographic Position Index (TPI), and Vector Ruggedness Measure (VRM). Additionally, we estimated the area of applicability (AOA) by presenting maps where the prediction model yielded high-quality results and where predictions were highly uncertain because machine learning (ML) models make predictions far beyond sampling locations without sampling data with no knowledge about these environments. The AOA method is well suited and unique for planning and decision-making about the best sampling strategy, most notably with limited data.