Monocular 6D pose estimation for objects is an essential but challenging task that is commonly applied in computer vision and robotics. Existing two-stage methods solve for rotations with Perspective-n-Point (PnP), which still incorporates translations, resulting in accuracy degeneration. In contrast, direct regression methods adopt Convolutional Neural Networks (CNNs) to solve for rotations and translations jointly but suffer from performance gaps in rotation accuracy. In this article, we propose a novel Normal Vector guided Regression Network (NVR-Net) to directly regress the 6D pose from a single RGB image under the guidance of 3D normal vectors. Specifically, we design a novel Orientation-Aware Feature (OAF) for pose estimation. It consists of two corresponding sets of 3D normal vectors to thoroughly disentangle rotation from translation estimation. Then, we introduce a CNN to predict a dense pixelwise representation of the OAF without viewpoint ambiguity. To estimate rotations and translations individually from the OAF, we propose a novel Pose from Normal Vectors (PNV) head networks under the instruction of a differentiable closed-form solution. Finally, extensive experiments on three common benchmarks demonstrate that our approach outperforms state-of-the-art methods on rotation accuracy and removes the gap between indirect and end-to-end methods. Moreover, our method can estimate the 6D pose of a single object within an RGB image in real-time.

The aircraft pose is the important measurement indicator related to the safe landing. The existing pose estimation methods are limited by the extraction accuracy of salient sparse points or line features, and it is hard to obtain higher accuracy and robustness, especially at long distance. In this paper, we implement attitude measurement utilizing a pose estimation. The aircraft pose estimation method based on dense features and pose consistency is proposed, which can infer the aircraft pose relative to the camera from the RGB image directly. This method uses the encoder-decoder network to predict the dense 3D coordinates, the surface regions and visible segmentation. Guided by the dense 2D-3D correspondences and surface regions, the aircraft pose is estimated using the pose regression module. To improve the robustness of decoupled pose estimation relative to object detection, this paper proposes a dual-channel regression framework connected with the pose consistency constraint, which enables mutual supervision between different dynamic zoom in (DZI). The experimental results on the aircraft dataset show the superiority of our aircraft pose estimation method significantly. The pose consistency constraints further improve the prediction performance on the aircraft dataset, LINEMOD dataset, and YCB-V dataset. The pose estimation method can be used for real-time measurement of aircraft attitude directly with the mean error of 0.377°, the RMS error of 0.491°, and the rate of 59FPS.

Attitude measurement is significant for airplane flight tests, especially in the landing phase. Among various technologies, off-board vision-based monocular measurement methods are flexible, non-intrusive, and self-contained. Previously employed model-based attitude measurement methods assume airplanes are rigid objects with fixed 3D models. However, airplane wing deformation caused by aerodynamic loads is neglected, which decreases the pose accuracy. To address it, this article proposes a novel method of estimating wing deformations and measuring airplane attitudes. It begins with building a polynomial wing deformation model using real wing data. Then, the wing deformation is reconstructed by leveraging dense correspondences between the input RGB image and the rendered image corresponding to a coarse pose estimation. After rectifying the wing deformation, a two-stage pose refinement strategy is proposed to refine the coarse pose. To evaluate our method, we propose a new airplane image dataset in which wing deformations are manually annotated. Experiments on this dataset demonstrate the outperformance of our method. Furthermore, results on airplane flight tests illustrate that our method can run on images of resolution 1920×1080 with a speed of 21 Hz and an accuracy of 0.4° for the three attitude angles.

Pose parameters constitute the fundamental flight state parameters of aircraft, providing a pivotal basis for aircraft design, performance evaluation, and flight safety assurance. The positional accuracy of vision measurements is often limited by the measurement distance. In response to the challenges associated with measuring the landing position of aircraft, the laser ranging position compensation method is proposed by combining a laser rangefinder (LRF) with the vision system to realize high-precision dynamic positioning of fixed-wing aircraft. Our method involves the construction of a laser-vision system to track and capture the aircraft images and the object distances. The aircraft attitude parameters are measured by combining a 6D initial pose estimation and the pose tracking method. In addition, leveraging the estimated 6D pose parameters and aircraft 3D model, the dense 3D coordinates map of the aircraft is rendered to locate the laser reflection point (LRP) on the aircraft’s surface and infer the origin of the aircraft coordinate frame accurately. By establishing the flight simulation scene, the position error effects caused by 6D pose errors, LR assembling errors, and laser ranging errors are simulated and analyzed numerically. The theoretical error propagation models are derived to corroborate the numerical simulation results. In a real flight test within a 3km landing range, the RMS errors of the aircraft attitude are below 0.5°, the standard deviations of the aircraft position errors are less than 1.5m, and the measurement rate exceeds 60fps, which demonstrates that the proposed method can be used for accurate real-time measurements of aircraft attitude and position.

The Antarctic region holds significant scientific research value and potential resources. Currently, limited research exists on the use of seismic exploration methods for Antarctic subglacial lakes compared to their use on other continents. Moreover, few reports are available on systems capable of multi-channel seismic data acquisition, remote data quality monitoring, and high-speed real-time data recycling in the extremely low temperatures of Antarctica. In this study, we developed a Zynq-based seismic acquisition station for polar exploration. The system features a compact design, lightweight construction, high data collection accuracy, excellent cold resistance, low power consumption, and real-time control. The software and hardware design of the system are described here, and validity testing is presented. The main controller utilizes a Zynq series system-on-chip integrated with an FPGA (Field-Programmable Gate Array) and an ARM (Advanced RISC Machine), enabling functions such as local data storage on a secure digital card, Wi-Fi wireless human–machine interaction, and high-speed Ethernet data transmission. Furthermore, to enhance data acquisition accuracy under low-temperature conditions, a neural network was employed for the temperature drift correction of the analog-to-digital converter chip. The validity test results showed that the station operated stably, was easy to use, and met the high-standard requirements for polar exploration.