Reflection-Robust Stereo Visual-Inertial Odometry via Kalman-Filtered Ground-Height Estimation
Reflection-Robust Stereo Visual-Inertial Odometry via Kalman-Filtered Ground-Height Estimation
This research presents a method for stereo visual-inertial odometry that addresses unreliable feature associations caused by specular reflections on indoor floors. Instead of suppressing all floor-region features, the proposed approach estimates ground height in real time using a scalar Kalman filter with Normalized Innovation Squared (NIS) gating and rejects only feature candidates whose inferred 3D positions fall below the estimated ground height within reflection-prone regions. Experiments on real reflective-floor indoor sequences demonstrate improved trajectory accuracy in reflection-dominant environments while maintaining accurate ground-height estimates and recovering from actual ground-height changes.
The paper introduces a lightweight front-end for stereo visual-inertial odometry that estimates ground height in real time using a Kalman filter to reject virtual features caused by specular reflections on indoor floors, improving trajectory accuracy in reflection-dominant environments while maintaining robustness to actual ground-height changes.
Source: arXiv Robotics — research abstracts · Read original article ↗
Loading article text…
What the source reports
Publisher-reported claims, with original evidence. These results have not been independently verified by RoboSignal.
- paper: not_reported
Source excerpts and review record
Automatically extracted; no manual editorial approval recorded.
nd pose uncertainty are used to construct frame-level ground-height measurements. A scalar Kalman filter with Normalized Innovation Squared (NIS) gating then provides a stable ground-height estimate, and an uncertainty-aware safety margin is used for selective below-ground virtual feature rejection. Experiments on real reflective-floor indoor sequences show that the proposed front-end generally improves trajectory ac
Open source S4
Source：arXiv Robotics — research abstracts · arxiv.org