Zheyu Liu
Contact information
Location: Space Lab Building 176
Thesis title
Development of novel graph-based visual SLAM systems in complex dynamic environments
Research overview
Simultaneous Localization and Mapping (SLAM) enables systems like robots or autonomous vehicles to map their environment and track their position without external positioning data (e.g., GPS). Graph-based SLAM uses a graph structure where nodes represent the robot’s poses (positions and orientations), and edges capture constraints between poses, derived from sensor data. These constraints are optimized to minimize errors, improving pose estimation accuracy. This project leverages graph theory to enhance visual SLAM by introducing novel constraints that account for dynamic environmental factors, such as moving objects and occlusions. These advancements aim to improve system robustness in complex and dynamic scenarios.