Identifiability-aware LiDAR-Inertial-BIM fusion method proposed

Yujie Zhang and other authors propose an identifiability-aware LiDAR-inertial-BIM fusion localization method for robot BIM localization and fault attribution under construction deviations. The approach reports attribution only when a fault is attributable; otherwise, it retains candidates, defers BIM updates and BIM-based pose corrections in ambiguous directions, and resumes once new references arrive. In a preregistered two-dimensional study, for detectable but non-attributable cases, the method's conditional error rate was 0.83%, compared with 31.67% for a baseline using uncertainty-based abstention at the same answer rate. In paired fault injections in a laboratory room (with its IFC model) and on three floors of the SLABIM dataset, the method produced no off-target attributions (0 of 168 conditions in the laboratory, 0 of 1302 conditions in SLABIM) and attributed all 8 laboratory faults predicted to be attributable; both reproduced baselines exhibited off-target structural markers in all settings.

AI 根据报道生成 · 2 days ago更新