Lidar and Neural Networks Revolutionize Concrete Inspection

Scientists at Drexel University have developed a groundbreaking solution that combines machine vision, laser scanning, lidar and neural networks to identify and measure structural damage in concrete. Using a convolutional neural network, the system detects cracks and evaluates their size with a laser scanner. Simultaneously, lidar captures the surrounding environment, enabling the creation of a detailed digital twin of the damaged structure.

This innovative approach enhances traditional inspection methods by offering a comprehensive, data-rich representation of structural damage. With a digital twin, building and architectural professionals can assess the severity of cracks and predict how they might evolve over time. This capability is invaluable for determining whether immediate intervention is necessary to prevent further deterioration or if periodic monitoring will suffice.

For example, organizations can rescan the same area over time and compare updated digital twins to track changes in structural integrity. The precision of lidar and 3D laser scanning provides insights that are nearly impossible to achieve through conventional visual inspections, leading to better-informed maintenance decisions and ultimately, more resilient infrastructure.

This robotic system illustrates how the integration of lidar and machine learning continues to expand the possibilities for structural health monitoring. By automating the process and generating actionable data, it reduces human error, saves time, and provides unparalleled accuracy.