SLABIM: A SLAM-BIM Coupled Dataset in HKUST Main Building (2502.16856v1)

Abstract: Existing indoor SLAM datasets primarily focus on robot sensing, often lacking building architectures. To address this gap, we design and construct the first dataset to couple the SLAM and BIM, named SLABIM. This dataset provides BIM and SLAM-oriented sensor data, both modeling a university building at HKUST. The as-designed BIM is decomposed and converted for ease of use. We employ a multi-sensor suite for multi-session data collection and mapping to obtain the as-built model. All the related data are timestamped and organized, enabling users to deploy and test effectively. Furthermore, we deploy advanced methods and report the experimental results on three tasks: registration, localization and semantic mapping, demonstrating the effectiveness and practicality of SLABIM. We make our dataset open-source at https://github.com/HKUST-Aerial-Robotics/SLABIM.

Overview of SLABIM: A SLAM-BIM Coupled Dataset

The integration of Simultaneous Localization and Mapping (SLAM) with Building Information Modeling (BIM) presents a potential leap forward in both mobile robotics and the architectural, engineering, and construction (AEC) industry. The paper "SLABIM: A SLAM-BIM Coupled Dataset in HKUST Main Building" introduces the SLABIM dataset, which is a novel contribution that aligns SLAM-oriented data with a richly detailed BIM model. This dataset is designed with the explicit aim of bridging the gap between robotic mapping technologies and digital building representations. By coupling these technologies within a publicly accessible dataset, the paper sets a foundation for advancing research in autonomous navigation and digital twin creation.

Dataset Composition and Acquisition

SLABIM is centered around a detailed BIM of the HKUST main building, spanning multiple floors and covering a wide variety of environments characteristic of indoor spaces. The dataset is constructed by decomposing the original BIM into manageable, reusable semantic components, thus making it adaptable for a multitude of practical applications. The dataset is enhanced with a comprehensive set of multi-sensor data gathered using a sophisticated sensor suite. This suite includes high-resolution fish-eye cameras, a LiDAR scanner, an IMU, and an RTK module, facilitating the collection of diverse SLAM-oriented data. Collected data is organized into 12 distinct sessions across various building levels, each capturing the complexity and variance of real-world indoor environments.

Key Applications of SLABIM

The usefulness of SLABIM extends to several critical applications in robotics and construction:

1. Global LiDAR-to-BIM Registration: The dataset supports global registration tasks necessary for aligning mobile robotic systems with digital building models. The paper tests contemporary methods, such as those based on branch-and-bound algorithms and pose hough transforms, indicating achievable precision and computational efficiency even amidst challenges posed by repetitive patterns and structural deviations typical in constructed environments.
2. Robot Pose Tracking on BIM: The dataset facilitates continuous pose tracking necessary for autonomous navigation, leveraging BIM-converted point cloud maps. Through an optimization-based approach, the researchers demonstrate effective pose estimation crucial for deploying mobile robots in complex indoor environments where GPS is unavailable.
3. Semantic Mapping Evaluation: SLABIM serves as a benchmark for evaluating semantic mapping algorithms by providing ground-truth annotations derived from BIM. This capability is crucial for assessing methods that leverage contextual scene understanding to improve navigation and interaction within indoor spaces.

Implications and Future Directions

The implications of the SLABIM dataset are multifaceted. Practically, it enhances the capabilities of mobile robots operating in indoor environments by providing a reliable method for localization and navigation. Theoretically, it opens avenues for the exploration of integrating complex digital models with real-time robotic data processing, promoting advances in digital twinning and autonomous building management systems. Furthermore, as BIM becomes increasingly integral to smart city infrastructures, the interoperability facilitated by datasets like SLABIM can drive innovation in construction automation and operational efficiencies.

Looking ahead, future research could explore the integration of more complex semantic data structures with SLABIM, enhancing its utility for applications demanding fine-grained spatial and procedural information. Additionally, the development of advanced machine learning models capable of handling the inherent noise and uncertainty in SLAM data could leverage SLABIM as a robust testing ground for iterative improvement. Expanding this dataset to include a broader spectrum of building types and environments would also offer substantial value, further solidifying SLABIM's role as a cornerstone resource in SLAM and BIM interdisciplinary research.