Low-cost Thermal Mapping for Concrete Heat Monitoring

Alex Junho Lee, Younggun Cho, Hyun Myung

TL;DR

This paper presents a low-cost mobile robotic system designed to create 3D thermal maps of concrete structures during the curing process. By fusing data from a 2D LiDAR, a consumer-grade IMU, and a thermal camera, the researchers provide a way to monitor concrete health and safety margins more accurately and affordably than traditional fixed-camera or air-temperature-based methods.

What Problem Does This Paper Address?

In the construction industry, the "curing" phase of concrete is a high-stakes period. If concrete dries too quickly or experiences extreme temperature gradients, it can develop cracks or fail to reach its intended structural strength. Currently, site managers often rely on ambient air temperature as a proxy for concrete heat, which is indirect and often inaccurate.

While thermal cameras can provide direct surface measurements, installing fixed cameras across a massive, evolving construction site is prohibitively expensive and impractical. There is a clear need for an autonomous, mobile solution that can navigate a site and generate a "digital twin" of its thermal state over time.

Methods

The authors developed a "Thermal-LiDAR SLAM" (Simultaneous Localization and Mapping) pipeline that prioritizes cost-effectiveness by using relatively inexpensive sensors:

  1. Sensor Fusion: The system uses a 2D LiDAR for distance measurements, a consumer-level IMU to detect the direction of gravity, and a 14-bit Long-Wave Infrared (LWIR) thermal camera.
  2. 2.5D Wall Reconstruction: Since a 2D LiDAR only scans a single horizontal plane, the authors assume the environment consists of vertical walls of a known height. By using the IMU to correct for robot tilt, they "extrude" the 2D scan lines vertically to create 3D wall surfaces.
  3. Thermal Unprojection: The 2D thermal images are projected onto these generated 3D walls. By knowing the camera's intrinsic parameters and its position relative to the LiDAR, the system assigns a temperature value to every point in the 3D cloud.
  4. Thermal-Enhanced Optimization: A key novelty is the use of Pose-Graph Optimization (PGO) that considers both geometry and temperature. When the robot revisits an area (loop closure), the system optimizes the map by minimizing the difference in both the position of the walls and their recorded temperatures.

Datasets

The researchers did not use standard public benchmarks. Instead, they collected a custom dataset at an active construction site in Heunghae-eup, Pohang, South Korea, in February 2020. The data was recorded using a mobile robot and stored in rosbag format for post-processing. The authors expressed intent to release this data in the future to aid the research community.

Key Insights and Ideas

  • Temperature as a SLAM Feature: Most SLAM algorithms rely purely on geometry or visual light features. This paper suggests that thermal intensity can act as an additional constraint, helping to refine the robot's pose estimation in environments where geometric features might be repetitive.
  • The 2.5D Shortcut: By leveraging the specific constraints of construction sites (mostly vertical walls and flat floors), the authors show that expensive 3D LiDARs aren't strictly necessary for meaningful structural monitoring.
  • Temporal Alignment: By using the Iterative Closest Point (ICP) algorithm to align thermal maps taken hours apart, the system can calculate the rate of temperature change, which is the most critical variable for predicting concrete strength and preventing defects.

Key Results

  • Successful Mapping: The system successfully reconstructed thermal maps of a construction site, clearly distinguishing between sun-facing walls (warmer/green) and shaded interior walls (cooler/blue).
  • Consistency: The SLAM pipeline remained stable across multiple "loops" or repetitions, with the PGO successfully aligning the maps without divergence.
  • Real-world Validation: The experiments confirmed that window frames and sunlight-exposed areas exhibited significantly higher temperatures, proving the system's sensitivity to environmental heat sources.
  • Monitoring Capability: The researchers demonstrated that they could monitor the overall temperature increase of a site over time, providing a quantitative basis for adjusting curing periods.

Limitations and Future Work

The primary limitation is the 2.5D assumption. Because the system assumes all detected points belong to vertical walls, it cannot accurately map floors, ceilings, or complex non-vertical structures. Furthermore, the current algorithm struggles to distinguish between permanent concrete walls and temporary obstacles like handrails or construction equipment.

The authors plan to expand this work into full 3D mapping. They suggest that by using thermal images to provide additional constraints, they can filter out "non-wall" objects and create a more robust, high-fidelity thermal digital twin of the entire construction environment.

Summary updated

Abstract

Robotics has been widely applied in smart construction for generating the digital twin or for autonomous inspection of construction sites. For example, for thermal inspection during concrete curing, continual monitoring of the concrete temperature is required to ensure concrete strength and to avoid cracks. However, buildings are typically too large to be monitored by installing fixed thermal cameras, and postprocessing is required to compute the accumulated heat of each measurement point. Thus, by using an autonomous monitoring system with the capability of long-term thermal mapping at a large construction site, both cost-effectiveness and a precise safety margin of the curing period estimation can be acquired. Therefore, this study proposes a low-cost thermal mapping system consisting of a 2D range scanner attached to a consumerlevel inertial measurement unit and a thermal camera for automated heat monitoring in construction using mobile robots.
BibTeX
@misc{lee2022low,
  author = {Alex Junho Lee and Younggun Cho and Hyun Myung},
  title = {{Low-cost Thermal Mapping for Concrete Heat Monitoring}},
  year = {2022},
  eprint = {2211.02244},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2211.02244},
}

Loading related content...

Figure