Research about navigation in our lab targets to devise systems that can perform SLAM(Simultaneous Localization and Mapping) and navigation in human-like ways. There are three main ongoing projects in this part: (a) line-based indoor SLAM that uses vanishing points as absolute direction indicators, floor and vertical lines as well as point features for its landmarks, (b) navigation imitating humans based on localization using view points, path integration and reorientation, (c) semantic SLAM that maps based on integration of topology and semantics, thus enables low-cost sensors’ performance maximization.

Line-based indoor SLAM
- SLAM using vanishing points, floor and vertical lines
- Lines and points hybrid SLAM


Navigation imitating humans
- Localization depending on view points
- Path integration, Reorientation


Place Recognition using Straight Lines for Vision-based SLAM

Outdoor Place Recognition in Urban Environments Using Straight Lines

Semantic SLAM
- Mapping by integrating topology and semantics
- Performance maximization of low-cost sensors



  2011.10.01~2012.03.31      Development of vision based SLAM system for indoor environments      Hanool Robotics