Sign in Subscribe
Machine learning

New Field Robotics Study Findings Recently Were Reported by Researchers at China West Normal University

Strategic Insights

— 2 min read

Researchers have proposed a direct geometrically constrained SLAM method based on target detection and depth image segmentation, named YGDD-SLAM. This method can work robustly, accurately, and continuously in highly dynamic environments. The system first acquires static and potential dynamic feature points in the current frame through a target detection network, and then identifies dynamic targets by combining the geometric change relationship between static and potential dynamic feature points between adjacent frames. The dynamic object regions at the pixel level are segmented out based on the double-peak feature of the gray-scale histogram of the dynamic target region in the depth image.

Key Takeaways:

  • The proposed YGDD-SLAM method can work robustly, accurately, and continuously in highly dynamic environments.
  • The system uses a target detection network to acquire static and potential dynamic feature points in the current frame.
  • Dynamic targets are identified by combining the geometric change relationship between static and potential dynamic feature points between adjacent frames.
  • The dynamic object regions at the pixel level are segmented out based on the double-peak feature of the gray-scale histogram of the dynamic target region in the depth image.
  • The YGDD-SLAM system is validated on TUM data set and Bonn data set, and it significantly improves the localization accuracy and system stability in different types of dynamic environments.
  • The research concluded that the YGDD-SLAM system is a more efficient and accurate SLAM method compared to existing methods.
  • The YGDD-SLAM method is applicable to various fields, including robotics and autonomous vehicles.
  • The research team, led by Zhengyong Feng, consists of Peng Liao, Liheng Chen, and Jialiang Tang.

Statistics:

  • The YGDD-SLAM method is validated on two data sets: TUM data set and Bonn data set (Source: Ygdd-slam: Direct Geometric Constraint Slam Based On Object Detection and Depth Image Segmentation. Journal of Field Robotics, 2025).
  • The localization accuracy of the YGDD-SLAM system is significantly improved in different types of dynamic environments (Source: Ygdd-slam: Direct Geometric Constraint Slam Based On Object Detection and Depth Image Segmentation. Journal of Field Robotics, 2025).
  • The YGDD-SLAM system can work robustly, accurately, and continuously in highly dynamic environments (Source: Ygdd-slam: Direct Geometric Constraint Slam Based On Object Detection and Depth Image Segmentation. Journal of Field Robotics, 2025).

Sources:

  • Ygdd-slam: Direct Geometric Constraint Slam Based On Object Detection and Depth Image Segmentation. Journal of Field Robotics, 2025.
  • Zhengyong Feng, China West Normal University, School of Electronic Information Engineering, Nanchong, People's Republic of China.
  • Peng Liao, Liheng Chen, and Jialiang Tang, China West Normal University.

Read more