Event
PinNet improves LiDAR loop closure detection with keypoint descriptors
1 reports · 1 sources
Yanlong Ma, Brett T. Lopez, and others proposed the PinNet neural network, which generates keypoints and their local geometric descriptors from point clouds for place recognition and inter-frame registration, and adds a plane-based geometric self-attention module to model spatial relationships among keypoints, thereby enhancing descriptor discriminability. The method was evaluated on datasets collected by various different LiDAR sensors; the authors claim it has strong place recognition performance, accurate relative pose estimation, and successful single-shot localization in different environments.
Report timeline
- PinNet: Keypoint-Aware Local Descriptors with Geometric Embedding for LiDAR SLAM Loop Closure Detection
PinNet generates point cloud keypoints and their descriptors, and uses a plane-based geometric self-attention module to model spatial relationships between keypoints, thereby improving descriptor discriminability for loop closure detection and point cloud registration. The method was evaluated on datasets collected with various different LiDAR sensors; experimental results show strong place recognition performance, precise relative pose estimation, and successful single-shot localization in different environments.
arXiv cs.RO ·