Abstract
In the event of a disaster, unmanned construction with remotely operated construction machinery is critical for quick disaster recovery. Those machines can weight up to several tons and can easily sink on inadequate soil. Therefore, it is important to judge the trafficability of remotely operated construction machinery at a disaster site. In this research, we propose a non-contact method for judging trafficability. The proposed method classifies the soil type and estimates the water content using spectral images. The cone index is then estimated and the trafficability is judged from the cone index. As experiment, we judged the trafficability for a real construction machine using the proposed method. The results showed the effectiveness of the proposed method based on soil type classification and water content estimation.
Introduction
Disaster recovery efforts often rely on heavy construction machinery to clear debris and stabilize terrain. However, in remote or hazardous environments, operating these multi-ton machines carries the risk of immobilization if the ground is too soft. Traditional methods for assessing soil strength, such as measuring the "cone index" (CI) with physical penetrometers, are slow and potentially dangerous for personnel in disaster zones. This paper presents a non-contact, remote sensing approach to evaluate terrain safety before machinery is deployed.
Proposed Methodology
The proposed system follows a structured three-step pipeline to determine if a specific piece of machinery can safely traverse a given area.
- Soil Classification: A multi-spectral camera captures data in the 350–1100 nm range. A neural network then identifies the soil type (e.g., clay, loam, sand), as different materials exhibit unique physical responses to moisture.
- Water Content Estimation: The system calculates the difference in reflectance between near-infrared (NIR) bands (900–1700 nm) and visible light (570 nm). This "spectral difference" provides a robust estimate of moisture levels across varying lighting conditions.
- Trafficability Judgment: By combining the soil type and estimated moisture, the system calculates the expected Cone Index. If this value exceeds a safety threshold (e.g., 200 kN/m²), the terrain is flagged as safe for operation.
Soil Parameter Estimation
A key innovation of this research is the use of spectral reflectance to predict physical soil properties. The relationship between moisture and reflectance is highly non-linear and varies significantly by soil type. For instance, clay and sand with identical water content will have vastly different load-bearing capacities. The researchers developed specific calibration curves for various soil types to translate spectral data into accurate moisture and strength estimates.
Experimental Results
The system was validated using a 13.4-ton hydraulic excavator in a controlled field test. The neural network achieved a high classification accuracy, with 97.3% precision in identifying soil types. Furthermore, the estimated cone index showed an error rate of only 5.26% for hard ground and 10.3% for soft ground. During field trials, the system successfully predicted areas where the excavator would sink, correctly identifying "soft ground" zones as untraffickable.
Conclusion and Future Work
The proposed non-contact method demonstrates high accuracy in predicting soil trafficability, offering a safer and faster alternative to manual measurements in disaster recovery. While effective, future work will focus on expanding the soil database to include more complex mixtures and integrating sensors that can account for subsurface conditions, ensuring the system isn't misled by dry surface crusts over deep mud.
Summary updated