A5003421323- Robotics and Sensor-Based Localization
- Planetary Science and Exploration
- Satellite Image Processing and Photogrammetry
- Advanced Vision and Imaging
- Inertial Sensor and Navigation
- Space Satellite Systems and Control
- Robotic Path Planning Algorithms
- Statistical and numerical algorithms
- Infrared Target Detection Methodologies
- Vehicle License Plate Recognition
- Advanced Image Processing Techniques
- Advanced Sensor and Control Systems
- Advanced Research in Science and Engineering
- Remote Sensing and LiDAR Applications
- Optical measurement and interference techniques
- 3D Surveying and Cultural Heritage
- Video Surveillance and Tracking Methods
- Image and Object Detection Techniques
- Industrial Vision Systems and Defect Detection
- Advanced Statistical Methods and Models
- CCD and CMOS Imaging Sensors
- Hydrocarbon exploration and reservoir analysis
- Image Processing and 3D Reconstruction
- Remote Sensing and Land Use
- Advanced Algorithms and Applications
Beijing Institute of Big Data Research
2022-2023
Aerospace Information Research Institute
2020-2023
Chinese Academy of Sciences
2015-2023
Liaoning Technical University
2023
Institute of Remote Sensing and Digital Earth
2015-2020
University of Chinese Academy of Sciences
2019
Wuhan University
2014
Fujian Institute of Oceanography
2014
Craters on the lunar surface are most direct method for study of geological processes and great significance to evolution. In order fill research gap small craters (diameter less than 3 m), we focus around moving path Yutu-2 rover carry out a 3D reconstruction geometrical morphology analysis them. First, self-calibration model with multiple feature constraints is used calibrate navigation camera obtain internal external parameters. Then, sequence images overlapping regions from neighboring...
Abstract Precise localisation of the Yutu‐2 moon rover plays an important role in path planning, obstacle avoidance and navigating to target features. To provide high‐precision information, a stereo bundle adjustment method using theory unit quaternions is presented for first time. improve precision robustness proposed method, rover's pose, from visual odometry technique assisted by inertial measurement rotation angles mast mechanism, viewed as pseudo‐observation. A reasonable weighting...
Currently, deep neural networks have gained impressive performance for object detection when plenty of labeled training data are available. However, a substantial quantity is typically rare and time-consuming to obtain in space exploration. To possibly alleviate this problem, we use domain adaptation (DA) effectively detect unannotated real samples the crater problem with help autoannotated synthetic only. Specifically, present novel network, namely, CraterDANet, which unifies image-...
In order to complete the high-precision calibration of planetary rover navigation camera using limited initial data in-orbit, we proposed a joint adjustment model with additional multiple constraints. Specifically, base was first established based on bundle model, second-order radial and tangential distortion parameters. Then, combining constraints collinearity, coplanarity, known distance relative pose invariance, constructed realize in orbit self-calibration camera. Given problem...
Realizing the pinpoint landing localization for Chang'E-3 system is a symbol of success this mission, and it also basic requirement lunar Lander Rover to carry out following scientific exploration tasks.In paper, jointly dealing with descent image sequence Chang'E-2 DOM data, an approach based on images processing techniques, such as SIFT feature matching, homography transportation stereo vision measurement, presented realize system.The result proved be reliable accuracy will critical...
The stereo vision system is the special engineering measurement instrument of Chang'e-4 lunar rover. It composed Navigation Camera (NavCam) and Mast Mechanism (MasMec). An improved self-calibration method for rover proposed. consists two parts: NavCam's MasMec's self-calibration. A combined adjustment based on points lines baseline constraint NavCam considered. model MasMec established product-of-exponentials formula. Finally, premission laboratory calibration on-site are carried out. shows...
High-precision localization technology is an important component for conducting scientific explorations in the Chang'E-3 lunar rover mission.In this study, exact stereo images strips network (SISNL) method proposed to solve problem of lack high-precision control points on surface.First, we obtained camera calibration parameters using bundle adjustment algorithm a ground field with some known points.Hence baseline and relative orientation elements cameras were acquired.Second, after building...
Aiming at the problems of poor recognition effect and low rate existing methods in process belt deviation detection, this paper proposes a real-time detection method. Firstly, ResNet18 combined with attention mechanism module is used as feature extraction network to enhance features edge region suppress other regions. Then, extracted are predict approximate locations edges using classifier based on contextual information fully connected layer. Next, improved gradient equation structural loss...
An exact lunar rover navigation and localization algorithm used stereo images is proposed in order to obtain hign-precision information. Firstly, A brief introduction of the given this paper, included coordinate transformation theory Chinas first rover, relative orientation based on images, absolute usedstereoscopic models weight matrix observations coordinates tie points. Then, procedure given. Last, Some conclusions are drawn after experiment results rover. The approved can improve...
In this paper, we present a descent image based extraterrestrial terrain reconstruction technology, which has been successfully applied in China's first soft landing Chang'E-3 mission on moon. When scouting the moon surface, lunar rover (Jade Rabbit rover) requires reliable information for navigation and positioning tasks. The sequence acts as link between orbital images surface images. proposed approach collects multilevel images, well inertial from an measurement unit stereo vision...
<p indent=0mm>At 7:18 on May 15, 2021, Beijing time, more than <sc>9 months</sc> after the launch of China’s first Mars exploration mission Tianwen-1, lander successfully landed in pre-selected landing area southern Utopia Plain Mars. The overall terrain this is relatively flat, making it easier for Tianwen-1 to land safely, but there are still scattered obstacles such as craters, sand dunes, and ridges. Therefore, needs be accurately located, order plan path rover, preventing rover from...
Abstract The hand–eye system is an important component of the Zhurong rover, which crucial for positioning, terrain reconstruction, stereo vision measurement and mission planning functions. consists a navigation camera mast mechanism, its positioning accuracy has effect on teleoperation rover. A method that combines close‐range photogrammetry robot kinematics proposed. It includes geometric calibration, coordinate transformations. Experiments were carried out in different scenes (for...
As a scientific payload of Chang’E-4 lunar rover, the panoramic camera not only detects geology but can be directly used for mapping and constructing terrain in rover’s navigation planning. We choose one model calibrating early; however, we often need to another convenience work afterward. This paper proposes distortion removal method based on parameter transformation algorithm, which overcomes mismatch problem between calibration model. The support points dense matching, spatial forward...
Tianwen-1 is the first Mars probe launched by China and mission in world to successfully complete three steps of exploration (orbiting, landing, roving) at one time. Based on unverifiable descent images which cover full range landing area, trajectory recovery fine terrain reconstruction are important parts planetary process. In this paper, a novel (TR-TR) algorithm employing proposed for dual-restrained conditions: restraints flat resulting an unstable solution parabolic causing low accuracy...
Abstract To set the sleeping mode for Yutu-2 rover, a visual pose prediction algorithm including terrain reconstruction and estimation was first studied. The precision is affected by using only stereo navigation camera (Navcam) images rotation angles of mast. However, hazard (Hazcam) fixed, an image network constructed linking all Navcam Hazcam stereoimages. Then, refined based on multiview block bundle adjustment. experimental results show that mean absolute errors check points in proposed...