Monte Carlo Localization (MCL) in Outdoor Environment
- Tae-Bum Kwon (Ph.D. )
- Yong-Hoon Ji (M.S.)
- Dong-Il Kim (Ph.D. Candidate, email@example.com)
- Research Background
- Reference map : DEM/DSM (built by an aerial mapping system)
- An elevation map cannot represent outdoor environments in detail.
- 3-D environments are observed differently from the air and the ground.
- Poor localization performance occurs when an elevation map is used as a reference map for outdoor environments.
Fig. 1 Examples of discrepancy between elevation map and real range sensor data
- COAG features : objects commonly observed from air and ground.
- COAG features can be accurately mapped into the elevation map built by an aerial mapping system and also sensed by a range sensor mounted on a mobile robot.
Fig. 2 Examples of COAG features
- Candidate Selection with Shape of Sensor Data
- Hausdorff distance (H(A, B))and average of minimum distances are used for classifying the candidates of robot.
Fig. 3 Example of Hausdorff distance and minimum distances
- Research Objectives
- Global localization and local tracking in outdoor environments
- Monte Carlo Localization(MCL) : determination of the robot pose based on Markov localization in a given map
Fig. 3 Experimental results of global localization
- Paper 1 :
Tae-Bum Kwon, Jae-Bok Song, A New Feature Commonly Observed from Air and Ground for Outdoor Localization with Elevation Map Built by Aerial Mapping System, Journal of Field Robotics, Vol. 28, No. 2, pp. 227-240, 2011.03.
- Paper 2 : Yong-Hoon Ji, Jae-Bok Song, Joo-Hyun Baek, and Jae-Kwan Ryu, Hausdorff Distance Matching for Elevation Map-based Global Localization of an Outdoor Mobile Robot, Journal of Institute of Control, Robotics and Systems, Vol. 17, No. 9, pp. 1-6, 2011.
- Paper 3 : Dong-Il Kim, Jae-Bok Song, Ji-Hoon Choi, Improvement of Localization Accuracy with COAG Features and Candidate Selection based on Shape of Sensor Data, Journal of Korea Robotics Society, Vol. 9, No. 2, pp. 117-123, 2014.
* Last updated: 2014. 7. 01