Evidence Alignment and Transfer Boundaries in 3D Point-Cloud Learning And Road-Scene Geometry

Abstract

This review examines a shared methodological problem in 3D point-cloud learning and road-scene geometry: how evidence from hierarchical spatial state-space aggregation for point-cloud classification can be placed in analytical dialogue with attention-based fusion of LiDAR and camera cues for curb detection without erasing differences in scale, assumptions, or intended use. The review draws on two focal records and 12 established sources already present in the project evidence cache. Its comparative framework links neighborhood construction, hierarchy, and state-space mixing to downstream questions of sampling robustness and efficiency. Comparison reveals recurring trade-offs among neighborhood construction, hierarchy, and state-space mixing. These trade-offs do not support a universal ranking; instead, they identify the operating envelope within which each method remains credible and the perturbations most likely to expose fragile conclusions. On this basis, the review proposes an auditable pathway from focal mechanism to application claim, with explicit checkpoints for calibration, external validity, and responsible interpretation.