A5114585385- Robotics and Sensor-Based Localization
- Advanced Image and Video Retrieval Techniques
- Advanced Vision and Imaging
- 3D Surveying and Cultural Heritage
- Advanced Neural Network Applications
- Robotic Path Planning Algorithms
- Metamaterials and Metasurfaces Applications
- Antenna Design and Analysis
- Advanced Antenna and Metasurface Technologies
- Evaluation and Optimization Models
- Evaluation Methods in Various Fields
- Advanced Optical Sensing Technologies
- Traffic Prediction and Management Techniques
- Image and Object Detection Techniques
Hefei University of Technology
2021-2024
Southeast University
2015
Lidar-assisted visual odometry is a widely used method for pose estimation. However, existing lidar-visual methods suffer from depth association errors, and single-point feature-based have insufficient accuracy are prone to tracking failures, leading inaccurate In this paper, we proposed direct monocular based on lidar fusion. Firstly, high-gradient pixels of the point cloud projection extracted, an initial estimation performed by minimizing photometric avoiding feature-depth associations....
A wideband dual-circularly polarized (dual-CP) magneto-electric (ME) dipole reflectarray with independent beam control is reported. dual-CP phase shifting unit cell proposed to provide manipulation on the two CP waves. For proof, a 2×2 bit operating at X-band 16×16 elements developed different directions. The results show that peak gain of prototype up 23.85 dBic 10.5 GHz an aperture efficiency 30.2%. joint bandwidth for 3-dB AR and around 22.5% (9.25-11.5 GHz). With advantages wide control,...
Simultaneous localization and mapping (SLAM) is one of the current research hotspots. However, in visual SLAM for dynamic environments, inaccurate detection object motion states incomplete region culling will lead to large errors. To address these issues, this paper proposes an RGB-D method based on feature association. The has strongly correlated features time space according input image sequence. Using moving probability points previous frame, movement frame calculated combination with...
In general, traffic demand regression analysis is conducted independently on a certain path of transport corridor with little consideration about the effects from other adjacent paths which have same directions. Separate multivariate linear and that multiple dependent variables are respectively based research data two between Shenyang Shanhaiguan. Analysis results indicate associative prediction has higher accuracy than separate prediction. Research this paper provide more accurate...
Most of the existing slam algorithms are designed based on assumption a static environment, this strong limits practical application most systems. The main reason is that moving objects will cause feature mismatch in pose estimation process, which turn affects accuracy localization and mapping. In paper, we propose SLAM algorithm dynamic environment. First, use BlendMask network to detect potential generate masks for objects. geometrically constrained joint optical flow method used points....