Opinion: How robots are learning to navigate the building site

Construction sites deny GPS through concrete, steel, and below-grade work, yet increasingly host autonomous robots. This article examines LiDAR-based Simultaneous Localisation and Mapping (SLAM), which lets a robot build and continuously update its own map in real time, suited to sites designed to change. It bounds SLAM's role carefully: it handles localisation and mapping, not placement precision or semantic meaning. Drawing on field deployments including Oxford's AutoInspect at Gatwick Station, it separates demonstrated results from claims about safety, productivity, and reduced rework. Barriers include dust, occlusion, and integration cost. Survey and progress monitoring are the likeliest early adoption points.