Towards BIM-based robot localization: a real-world case study

Abstract

Conventional mobile robots rely on pre-built point cloud maps for online localization. These map points are generally built using specialized mapping techniques, which involve high labor and computational costs. While in the architectural, engineering and construction (AEC) industry, asplanned building information modelings (BIM) are available for management and operation. In this paper, we consider the use of the digital representations of BIM for robot localization in built environments. First, we convert BIM data into localization-oriented point clouds, which is easy to implement and operate compared to relatively complex SLAM systems. Then, we perform iterative closest point (ICP)-based localization on the metric map using a laser scanner. The experiments are tested using collected laser data and BIM in the real world. The results show that ICP-based localization can track the robot pose with low errors (< [0.20m, 2.50°]), thus demonstrating the feasibility of BIM-based robot localization. In addition, we also discuss the reasons for errors, including the deviations between as-planned BIM and asbuilt status.