DBNtech

Major Object Detection and Matching System in Moon Local Video

Summary

We are conducting research to derive meaningful data by analyzing camera images of robots (rover) that will be used in the future lunar base construction as a consignment research from the Korea Institute of Civil Engineering and Building Technology (KICT). To build a lunar base, it is crucial to quantify local geospatial information and to monitor the construction processes through images taken from the rovers. For achieve these, we need a technology that automatically quantifies the spatial information of main objects. We have developed an AI Model that recognizes the main objects in the images collected at the virtual lunar base construction site and determines whether they are the same as those collected from the stereo camera. We plan to develop a technology that calculates the relative coordinates of the rovers at the lunar surface and a technology that matches objects in real time for future works.

Capability

This object detection and matching technology can be used in various fields, such as face detection, space construction, etc.

Examples of detection: face detection, ultra-precise object tracking (people, vehicles, etc.), object data linkage between multiple cameras, etc.

Object Extraction

Left Image of Rover

Right Image of Rover

Object Matching

Visualization of Object Matching

Left Image of Rover

Right Image of Rover

Deep Leaning based Automatic Detection System of Crater on the Lunar Surface

Summary

Detecting craters on the lunar surface is one of the most important research areas in aerospace. Traditionally, this has relied on the expert's eyes with high-resolution Digital Elevation Model (DEM) images. However, since experts are humans, detection results are not consistent, depending on who does it. To solve this problem, we started a consignment research project of the Korea Institute of Civil Engineering and Building Technology (KICT) for the development of automatic lunar crater detection system based on Deep Learning. Using this system, we can detect large and small craters more precisely, and also can detect many craters which are hard to find with the naked eyes. Through this research, we have secured technology for terrain object image processing and automatic statistics in the field of space construction. In addition, this technology can be used in other fields such as construction and transportation infrastructure using deep learning-based object detection. In the future, we will be able to contribute to the creation of new markets related to the activation of extreme environment construction technology and future space exploration.

Capability

The object detection technology based on Deep Learning can be used in various fields according to the problem to be solved.

Examples of applications: visual inspection of facility images (bridges, buildings, etc.), product quality inspection (crack, corrosion, etc.), store product identification, etc.

Object Detection in Deep Learning Models

Existing DEM Image

DEM Image with
Shade Relief

Ground-truth

Conclusion

Liquefaction Modeling and 3D Analysis Module Based on Drilling Information

Summary

Liquefaction is a phenomenon in which soil loses its stiffness and shear strength when subjected to stress and becomes liquid. In general, when liquefaction occurs, some of the water and sand are ejected onto the ground and create a space, leading to a risk of road sinking or sinkholes. Therefore, there is a growing need for liquefaction risk map production in Korea. In this study, we extracted data related to national drilling information for liquefaction risk map production; based on this information, we developed a module for data modeling and 3D analysis. This is the liquefaction damage prediction system developed jointly with the Korea Institute of Civil Engineering and Building Technology (KICT). Dividing the liquefaction risk of each region by grade from national drilling data provided by the Integration Database Center of the National Land and Ground Information, we can finally create a 2D and 3D-based liquefaction risk map.

Capability

This technology can be used to establish a system for linking with other systems such as earthquake disaster response and underground safety management; this technology can be used to build a comprehensive decision-making support system, such as safe construction and earthquake disaster response. This technology can be developed into a ground depression (sinkhole) risk prediction model based on deep learning algorithms in the future.

Ground Depression Prediction System

Summary

Ground depression is a phenomenon caused by the melting of underground rocks or the collapse of an existing cave. There are several main causes of it in urban areas: unstable construction, damage to aging water supply and sewer pipes, and inaccurate ground investigation and over-excavation during excavation work. In Korea, after the large-scale ground depression incident in Seoul in 2014, full-scale ground depression management measures have been established and implemented. Since most ground depressions have no signs, it is not known how and when they will occur. Therefore, for prevention and proactive response, a ground depression impact analysis system must be prepared. We have built a solution for a ground depression risk map and a GIS mapping system based on AI Deep Learning technology to provide a highly reliable ground depression risk map.

Capability

This technology can be used to analyze the impact of ground depression, analyze the reliability of extracted data, or use AI algorithms to apply data analysis to GIS mapping of ground depression risk maps. Through this, we can preemptively respond to the road damage caused by water leakage or soil loss caused by damage to the water and sewage system.

Mobile-based Integrated Underground Geospatial Information Map Service

Summary

With the recent urban development, the infrastructure for urban operation is becoming larger, including underground facilities and structures such as electricity, water and sewage, gas, heating, etc. However, the underground space integration map is neither up to date nor accurate. So, there is always a risk of large-scale accidents due to excavation, burial, or maintenance work of underground facilities. We have built an on-site utilization support system based on the latest and accurate integrated underground geospatial information map. In addition, we are developing a mobile-based integrated underground geospatial information map service to enhance its utility in various fields such as underground safety management and construction. These include mobile client modules, data security management, and underground map mobile visualization technology. Through these technologies, we have provided solutions to prevent underground safety accidents that may occur at excavation and burial sites in advance, and to respond to accidents systematically.

Capability

With this technology, we can easily check a large-scale underground space integration map at the construction site through mobile. This prevents accidents in underground spaces caused by excavation, etc., and enables systematic response in case of accidents in underground spaces.