Novel Non-feature Based SLAM: Joint Optimization of Non-feature Maps and Robot Trajectories
- Publication Type:
- Thesis
- Issue Date:
- 2025
Open Access
Copyright Clearance Process
- Recently Added
- In Progress
- Open Access
This item is open access.
This thesis investigates methods for joint optimization of robot poses and dense non-feature-based map representations in simultaneous localization and mapping (SLAM). Traditional non-feature-based SLAM approaches typically estimate poses and construct maps separately, which can limit consistency and accuracy. To address this limitation, the thesis proposes three optimization frameworks that couple localization with non-feature map estimation, an aspect largely unexplored in prior work. The first framework jointly optimizes robot poses and occupancy grid maps. The second develops a grid-based submap joining approach that performs joint optimization over submap poses and a global occupancy map, enabling scalability to large environments. The third extends joint optimization to signed distance field representations by formulating geometric constraints that update both robot poses and a volumetric TSDF map simultaneously. Together, these methods show that integrating pose estimation with non-feature-based mapping in a tightly coupled optimization process can significantly enhance a robot’s perception and spatial understanding.
Please use this identifier to cite or link to this item: