Track genomics, neuroscience and molecular biology.
Search papers, labs, and topics across Lattice.
GlassGuard addresses the critical challenge of LiDAR-based navigation failures caused by transparent and specular surfaces by reconstructing planar glass structures. The framework integrates a foundation vision model for instance segmentation with 3D structural cues and projective geometry for robust plane hypothesis generation and verification. Evaluated across nine building-scale scenes, GlassGuard achieves high glass coverage (85% panoramic, 82% pinhole) and significantly reduces free-space contamination (5-17x fewer false voxels) compared to baselines, enabling safer robot navigation.
Robots can now reliably navigate through buildings with glass, avoiding collisions and phantom obstacles by accurately mapping transparent surfaces.
Transparent and specular surfaces pose a serious challenge to LiDAR-based SLAM and navigation because laser returns may pass through glass, leaving collision boundaries absent from the map. Prior work attempts to reconstruct the missing surfaces, but inaccurate obstacle placement can create the opposite failure: contamination of traversable free space. Recognizing this dual requirement, we present GlassGuard, a navigation-oriented framework for reconstructing planar architectural glass from complementary visual and LiDAR evidence. We formulate success in terms of both glass coverage and free-space contamination and apply this principle throughout proposal verification and global map construction. A foundation vision model provides glass-instance masks, structural 3D cues generate metric plane hypotheses, and depth-free 2D projective geometry checks their orientations before they enter a consolidated global map. We evaluate GlassGuard in nine building-scale scenes spanning diverse glass structures, spatial scales, and lighting conditions, with more than one hour and 2.1 km of real-world robot traversal. GlassGuard achieves 85% of total glass coverage for its panoramic version. Under identical pinhole inputs, GlassGuard achieves 82% total coverage, compared with at most 61% for the evaluated baselines, while producing 5-17x fewer false voxels per frame. Qualitative examples with a navigation planner illustrate the reconstructed planes blocking paths through glass while leaving traversable routes open. The project page is available at https://glassguardproject.github.io/.
CMU ML