Research Interests
Mobile Robotics (Exploration and mapping (SLAM), with focus on topological worlds)
Algorithms and complexity. Graph Theory. Bayesian framework
Software design and development.
Image processing.
Web/Internet technology.
Selected Projects
Exploring the minmial information in topological SLAM.
What does it take to do topological SLAM deterministically?
We explore the fundamental limits to topological SLAM, in terms of markering aids? We develop algorithm that shows that while a single undirected immovalbe marker is not sufficient to solve SLAM deterministically, a directional marker is. Such a marker provides unique location and orientation information in one vertex of the graph-like world. We claim that this is the miminal information required to do topological SLAM deterministically.
Rigid proof is given, cost bound is derived.
Topological SLAM with real robot platform.
Can we do topological SLAM using real robot system?
- set up and calibrated omni-directional sensing systems on an existing robotic system.
- developed calibration (undistoration) approach for the onmi-directional camera.
- develop Bayesian framework and algorithms for hallway following and junction detection and classification (python, Java, C, C++).
On-going work: develop software platform on ROS (Robot Operating System),