#South Korea #construction technology #Incheon #excavator tracking #Incheon National

Enhanced Excavator Tracking for Construction Sites

In a groundbreaking study published in Automation in Construction, researchers from Incheon National University in South Korea have unveiled a revolutionary method for excavator tracking in dynamic construction environments. This innovative approach addresses one of the most pressing challenges in the industry: the frequent occlusions caused by overlapping activities on site, which often degrade the performance of traditional tracking systems.

The Challenge of Occlusions

Occlusions occur in construction settings when multiple machines interact closely, blocking the view of cameras used for monitoring. Traditional tracking methods struggle to maintain accuracy under these conditions, often assuming that at least one camera has an unobstructed view. However, reality shows that these ideal conditions are rare; hence, tracking often falters when critical information is obstructed. To remedy this issue, the research team led by Professor Choongwan Koo developed a reliability-based multi-camera strategy, utilizing deep learning for instance segmentation to enhance tracking amid simultaneous occlusions.

A Multilayered Approach

The researchers' approach employs an automated system that quantitatively evaluates and selects the most reliable camera feed in real-time. By analyzing occlusion ratios and identifying critical occlusion thresholds—for the excavator arm (0.7) and the body (0.5)—this strategy allows for a more nuanced understanding of which camera provides the clearest view at any given moment. This not only improves the accuracy of the tracking data but also establishes a benchmark for construction teams to assess their visual monitoring data quality, paving the way for improved decision-making on-site.

Practical and Economic Relevance

Beyond its methodological advancements, the proposed system holds significant practical implications. Enhanced tracking reliability directly translates into better operational logs for equipment, a crucial component for assessing carbon emissions and adhering to regulatory reporting requirements. This systemic improvement can mitigate administrative burdens and diminish the financial risks associated with penalties or unexpected costs due to revalidation needs. Importantly, the design leverages existing camera infrastructure, negating the need for extensive new installations and minimizing the complexities associated with power, networking, and site restrictions.

Reducing Costs and Labor Dependence

Additionally, automating the camera selection process lessens the dependency on manual oversight, thus reducing labor costs and operational inefficiencies. This multifaceted strategy not only simplifies the monitoring processes but also provides a scalable solution to enhance the reliability of vision-based construction systems. As a generalizable framework, the findings could easily extend to areas such as productivity analysis, activity recognition, and carbon emission monitoring, delivering far-reaching benefits across the construction industry.

Conclusion

The research presents a significant leap forward in construction technology, offering a comprehensive, automated solution for excavator tracking in environments laden with challenges. As construction continues to evolve, advancements like these will play a crucial role in optimizing operations and enhancing efficiency in the industry.

Reference

Title of original paper: Automated reliability-based multi-camera strategy for excavator tracking under dynamic occlusion using deep learning with instance segmentation.
Journal: Automation in Construction.
DOI: 10.1016/j.autcon.2025.106589

For further details, refer to the official website of Incheon National University here.