KLTNet: Learning-Based Sparse Feature Tracking for Improved Monocular VIO
zhenjun_zhao · x · 2026-08-26
The paper presents KLTNet, a learning-based sparse feature tracker designed to replace classical KLT trackers in VIO front ends. It uses a coarse-to-fine, dense-to-sparse architecture, combining low-resolution dense optical flow for robust initialization with triplet-patch refinement for accuracy. Experiments on VINS-Mono and OpenVINS show improved tracking and odometry accuracy while maintaining real-time performance on embedded platforms.
More from Embodied
- China's Robot Games Serve as Real-Life Labs for Humanoid Testing — CurieuxExplorer · 2026-08-26
- LightNav-0: A Generalist Navigation Brain for Any Robot — teortaxesTex · 2026-08-26
- Humanoid robot 'ChatGPT moment' is about autonomy, not human-level mimicry — VraserX · 2026-08-26
- Yuequan Bionic Unveils 38-DOF Bionic Robotic Hand — teortaxesTex · 2026-08-26
- CoRL 2026 Agentic Robotics Workshop Calls for Demos and Papers — GuanyaShi · 2026-08-26
- KUKA Robot Demo: Keeping Tool Center Point Still — Olivier__OG · 2026-08-26