Search papers, labs, and topics across Lattice.
RRTrack is a novel 6D pose tracking system designed to operate effectively in dynamic and occluded environments by integrating a memory-based video object segmentation approach with a dual-bank template matching module. This method allows for robust tracking even during fast motion and target disappearance, significantly improving performance metrics such as equal-subset mean ADD-S AR and ADD-S AUC by over 65% compared to existing methods. Experimental validation in both synthetic benchmarks and real-world scenarios confirms RRTrack's capability to maintain real-time efficiency at 55.2 FPS while ensuring accurate pose tracking under challenging conditions.
RRTrack achieves a 66.3% improvement in pose tracking accuracy, even in scenarios of complete occlusion and rapid motion.
Robust object 6D pose tracking is critical for robotic systems operating in dynamic and occluded scenes. Per-frame estimators are accurate but computationally expensive, while current trackers struggle with fast motion and complete occlusion due to their reliance on continuous visibility. To address these challenges, we present RRTrack, an efficient, recoverable object 6D pose tracker that enables robust tracking through fast motion and target disappearance--reappearance. RRTrack introduces a 2D--6D closed-loop tracking strategy that integrates memory-based video object segmentation (VOS) with 6D pose refinement. The 2D branch maintains target localization, and the 6D branch verifies geometric consistency before memory updates. In addition, a DINOv2-based dual-bank template matching module is developed to recover lost targets by jointly exploiting offline synthetic templates and online observation anchors while maintaining real-time efficiency. We also introduce a synthetic RGB-D benchmark comprising three robotic scenarios with fast motion and full occlusion. Experimental results on the synthetic benchmark demonstrate that RRTrack improves equal-subset mean ADD-S AR by 66.3\% and ADD-S AUC by 65.7\% over FoundationPose while achieving 55.2 FPS. Real-world experiments further validate the robustness of RRTrack under noisy sensing conditions. Project page: https://github.com/7kevin24/RRTrack
