Multi-Session Mapping for Indoor Substation Environment Using a Head-Mounted RGB-D Sensor

Three-dimensional model of environment combined with workflow information can provide intuitive and effective guidance for wearable-assisted maintenance and inspection in substations. Incremental mapping and sub-map stitching are essential for building a large-scale substation environment. In this paper, a multi-session mapping method using head-mounted RGB-D sensor is introduced for simultaneous localization and mapping (SLAM) in unknown substation environment. A hierarchical map structure is designed to store the sub-map models of the environment and equipment information, which allows further hierarchical query and visual display. Transformations among sub-maps are calculated based on cross-sub-map loop closure detection and ICP algorithm, followed by sub-maps alignment and stitching. A wearable visual SLAM system is developed, which integrates dense-mapping of electrical equipment as well as semi-dense mapping of large-scale substation environment. The result of experiment verifies the effectiveness and practicability of the proposed method.