Damage detection system with an artificial intelligence–supported unmanned aerial vehicle for use in disaster management

Abstract

The invention relates to a damage detection system including an artificial intelligence–supported unmanned aerial vehicle enabling high-accuracy damage detection in enclosed, complex, and narrow areas. This system is equipped with a 2D LIDAR, an RGB-D camera, and ultrasonic sensors, and the data collected from the sensors on the UAV are combined using fusion algorithms to create a high-accuracy three-dimensional model of the environment. In this way, structural weaknesses and hazardous areas in disaster zones can be identified and presented to response teams in real time.

Description

[0001] DESCRIPTION

[0002] DAMAGE DETECTION SYSTEM WITH AN ARTIFICIAL INTELLIGENCE- SUPPORTED UNMANNED AERIAL VEHICLE FOR USE IN DISASTER MANAGEMENT

[0003] TECHNICAL FIELD

[0004] The invention relates to a damage detection system including an artificial intelligence-supported unmanned aerial vehicle enabling high-accuracy damage detection in enclosed, complex, and narrow areas. This system is equipped with a 2D LIDAR, an RGB-D camera, and ultrasonic sensors, and the data collected from the sensors on the UAV are combined using fusion algorithms to create a high-accuracy three-dimensional model of the environment. In this way, structural weaknesses and hazardous areas in disaster zones can be identified and presented to response teams in real time.

[0005] PRIOR ART

[0006] Today, traditional methods used in damage detection for disaster management are based on manual inspections and images obtained from fixed cameras. Although UAV and computer vision technologies are used in post-disaster damage detection, UAVs that can perform precise maneuvers in enclosed areas and create a 3D model with high accuracy are limited.

[0007] Today, these technologies are rapidly developing for rapid data collection and damage detection after disasters. Some current applications in this technical field: Computer Vision and Artificial Intelligence-Supported Damage Detection: Within the scope of the Stanford University Disaster Image Analysis Project; deep learning algorithms are used to detect damage occurring in buildings after natural disasters via Al-based image analysis. The performance of models such as CNN and YOLO on different disaster types has been evaluated.

[0008] GEM (Global Earthquake Model) Damage Assessment Platform: GEM provides a comprehensive solution for damage detection by using Al and data analytics to assess building damage after earthquakes worldwide. By automatically classifying building damage with Al algorithms, it enables rescue teams to focus on priority areas.UAV-Based Damage Detection Systems: DJI has developed UAVs with high-resolution imaging and mapping features for use in disaster zones. In particular, DJI’s Mavic series is integrated with various software and is used in mapping post-disaster building damage.

[0009] SenseFly’s eBee: senseFly’s eBee UAV is used for damage detection in large areas. Rapid scanning of disaster zones is achieved with high-resolution cameras integrated into the UAV and GIS software.

[0010] SLAM-Based 3D Modeling Systems: Pix4Dmapper, a software used for 3D modeling, is a SLAM-based application capable of 3D mapping and modeling in order to accelerate damage detection in disaster zones. Images obtained from the UAV are processed in the software to create detailed indoor and outdoor building models. Other Sources and Application Examples: Google Earth Engine is a large satellite image database used to compare pre- and post-disaster images of disaster zones. This system is integrated with Al algorithms to provide data for post-disaster damage detection and analyses.

[0011] Search-and-Rescue and UAV-Supported Platforms: In various search-and-rescue platforms, UAV-based damage detection and building scans are performed, and they are used in disaster response processes especially by civil defense organizations in Europe and America.

[0012] UAV-supported damage detection systems in the prior art cannot provide precise navigation in enclosed areas due to limited GPS access. This is a serious obstacle to unmanned aerial vehicles entering enclosed areas and collecting data for damage detection. Moreover, UAVs lack the hardware and capabilities required for high-resolution indoor 3D modeling. It is not possible to fully perform damage detection with detections made from outside the building. The rapid and secure analysis of data obtained after a disaster cannot be carried out with sufficient effectiveness in existing prior-art systems. Since there are times after a disaster when organized work becomes difficult, although it is important for data to be collected, interpreted, and reported quickly and reliably, this cannot be met with prior-art systems. The lack of collision avoidance mechanisms in unmanned aerial vehicles restricts safe flight in complex indoor environments. Due to the complexity of areas where post-disaster damagedetection will be performed and the limited technical possibilities, these deficiencies make damage detection quite difficult.

[0013] LIST OF FIGURES

[0014] Figure 1. General View

[0015] Figure 2. System Flow Diagram

[0016] Correspondence of Numbers Shown in the Figures

[0017] 1. UAV

[0018] 1.1 Battery

[0019] 1.2 Flight Control Board

[0020] 1.3 Telemetry

[0021] 2. RGB Camera

[0022] 3. 2D Lidar Sensor

[0023] 4. Ultrasonic Sensor

[0024] 5. GPS

[0025] 6. Programming Board

[0026] DETAILED DESCRIPTION OF THE INVENTION

[0027] The invention consists of the parts and sections: UAV (1 ) including battery (1.1), flight control board (1.2), telemetry (1.3), RGB camera (2), 2D lidar sensor (3), ultrasonic sensor (4), GPS (5) and programming board (6).

[0028] In disaster management, being able to perform fast, reliable, and high-accuracy damage detection — especially in enclosed and complex areas — emerges as a major challenge. In traditional methods in the prior art, since GPS access is limited in enclosed areas, it becomes difficult for UAVs to collect data safely and effectively. In addition, systems capable of high-resolution 3D modeling and providing detailed and integrated data are limited. This situation causes disruptions in rapid response and rescue processes after a disaster.

[0029] Our invention provides an integrated UAV (1) system enabling high-accuracy damage detection in enclosed and complex areas. This system is equipped with a 2D LIDAR sensor (3), an RGB-D camera (2), and ultrasonic sensors (4), and the data collected from the sensors on the UAV (1) are combined using fusion algorithmslocated on the programming board (6) and a high-accuracy three-dimensional model of the environment is created. In this way, structural weaknesses and hazardous areas in disaster zones can be identified and presented to response teams in real time.

[0030] The invention is characterized by sensor fusion, autonomous navigation, and Al-supported analysis units. By combining the data received from the 2D LIDAR sensor (3), the RGB-D camera (2), and the ultrasonic sensors (4) on the programming board (6), high-accuracy indoor and outdoor 3D modeling can be performed. An autonomous navigation system is provided that enables safe and independent movement in enclosed areas without GPS support. In this way, safe flight and detailed damage detection are enabled in enclosed areas where GPS signals weaken. All collected data are combined with fusion algorithms to create high-accuracy three-dimensional models in which damaged regions are determined. Thus, by analyzing the three-dimensional models, experts perform fast and reliable damage detection, and the post-disaster response process is supported.

[0031] Our invention makes an important contribution to disaster management by collecting faster, more comprehensive, and more reliable data in areas where traditional methods are insufficient. In order to perform high-accuracy damage detection in enclosed and complex areas, detailed three-dimensional models are created by combining LIDAR, RGB-D camera, and ultrasonic sensor data with a fusion algorithm. Thanks to the autonomous navigation feature, the unmanned aerial vehicle can move safely and collect data in complex and narrow areas formed after a disaster. By providing real-time data with Al-supported analysis, it provides the opportunity for rapid response, which is of vital importance after a disaster. Compared to traditional methods, faster, more reliable, and more comprehensive data are collected from disaster zones. Instant information is provided to response teams about structural weaknesses and hazardous areas, thus offering a safer working opportunity.

Claims

CLAIMS1. A damage detection system including an artificial intelligence-supported unmanned aerial vehicle enabling high-accuracy damage detection in enclosed, complex and narrow areas, characterized by;- 2D LIDAR sensor (3), an RGB-D camera (2) and ultrasonic sensors (4) that collect the data enabling high-accuracy indoor and outdoor 3D modeling,- Programming board (6) including Al-supported analysis units comprising autonomous navigation and artificial intelligence algorithms enabling safe and independent movement without GPS support.

Images