A feed-forward SLAM system jointly optimizes two-view odometry and multi-view submap constraints in a single Sim(3) factor graph, achieving 0.032 m ATE on TUM RGB-D and 0.020 m on 7-Scenes in the uncalibrated setting.
In: IEEE Conf
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
UniSim-SLAM: Feed-Forward SLAM with Unified Sim(3) Optimization
A feed-forward SLAM system jointly optimizes two-view odometry and multi-view submap constraints in a single Sim(3) factor graph, achieving 0.032 m ATE on TUM RGB-D and 0.020 m on 7-Scenes in the uncalibrated setting.