Range-Only Beacon Localization

Summary

Many Autonomous Underwater Vehicle (AUV) systems determine their position using a set of transponder beacons whose location is known in advance. We have developed a system capable of navigating even when beacon locations are completely unknown and have tested our algorithm on real data.

Outlier rejection is a major issue, and we have developed a powerful method of rejecting outliers that is not dependent on a prior. We cast the problem as a graph partitioning problem on a graph constructed from pair-wise measurement consistencies.

This work was presented at the IEEE AUV 2004 conference in June, 2004.

The outlier rejection portion of this paper has subsequently been expanded and enhanced in our paper "Single Cluster Graph Partitioning for Robotics Applications".

IEEE AUV 2004 Conference Materials

• Robust Range-Only Beacon Localization (pdf)

  @inproceedings{Olson2004,
  	author="Edwin Olson and John Leonard and Seth Teller",
  	title="Robust Range-Only Beacon Localization",
  	booktitle="Proceedings of Autonomous Underwater Vehicles, 2004",
  	journal="IEEE AUV",
  	year={2004}
  }
  

• Power point presentation (ppt)
• Navigation filter animation [quicktime (2.6M), MPEG4 (3.6M)]
• Beacon solution estimation animation [quicktime (2.6M), MPEG4 (5.9M)]

Additional Materials

• Outlier rejection results
  1. Unprocessed (raw) ranges [png]
  2. Processed ranges, inliers are circles [png]
  3. Processed ranges, only inliers [png]
• SLAM filter results on GOATS'04 data [quicktime (300K)]
• SLAM filter results, final (causal) results: [png]

People

Robotics, Vision, and Sensor Networks Group 32 Vassar Street, 32-33x Cambridge, MA 02139 Tel: 617-253-6583 Fax: 617-258-7413