History helps fix mistakes in repeated visual localization poses
Recovering from Pose-Selection Failures in Repeated Visual Localization
Computer Vision and Pattern RecognitionRobotics
Summary
Visual localization systems sometimes pick the wrong position because similar-looking spots confuse them. The authors propose History-Guided Pose Reselection (HPR), which looks at past location guesses to pick the better current position guess without changing it. Their method improves accuracy across many tested setups and works fast, even fixing nearly all previous mistakes on a challenging dataset. It works best when location guesses are frequent and does not need odometry.
What this means in practice
- •For robotics engineers: Improve repeated visual localization accuracy in robots by filtering pose candidates using historical pose forecasts to avoid errors from repetitive structures.
- •For drone developers: Enhance drone navigation reliability by integrating history-guided pose reselection to correct pose estimation failures in visually repetitive environments.
Authors
Jaehyun Kim, Wonseok Kang, Tae-Wan Kim
Abstract
Repetitive structures can give an incorrect pose candidate more inlier support than an accurate one, causing visual localization to select it. History-Guided Pose Reselection (HPR) uses coarse position forecasts from up to five past localization outputs to distinguish well-separated current-query candidates. The best-fitting forecast proposes an alternative; HPR accepts it only if every well-fitting forecast is much closer to it than to the localizer's pose. HPR needs no odometry and returns the selected candidate unchanged, using prediction for selection rather than pose refinement. With one development-set threshold, HPR improves recall within 5 meters and 10 degrees in all 12 main configurations of Hierarchical Localization (HLoc), MASt3R-based pose generation, and JIST-based retrieval, losing successes on fewer than 0.4 percent of queries. On NAVER Gangnam B1, HPR corrects 97 HLoc failures and loses six successes; gains persist within 0.25 meters and 2 degrees on all three HLoc test sets. Median HPR runtime on HLoc candidates is below one millisecond, excluding pose generation. Recovery weakens under less frequent localization.