Advanced Bird Wildlife Management System with Deep Learning-Based UAV Swarming and Acoustic Suppression for Airport Safety

Abstract

The present invention relates to an advanced bird and wildlife management system through deep learning-based unmanned aerial vehicle swarming and sonic suppression for airport safety. More specifically, it relates to an advanced bird and wildlife management system through deep learning-based unmanned aerial vehicle swarming and sonic suppression for airport safety that maintains the safety of aircraft and airports to avoid bird strikes and utilizes unmanned aerial vehicle (UAV) swarming technology equipped with real-time bird detection capabilities using deep learning models and sonic deterrents to solve the critical problems of bird detection and wildlife management near airports by utilizing unmanned aerial vehicle (UAV) swarming and ultimately enhances the safety of both aircraft and bird species. Accordingly, the present invention provides an advanced bird and wildlife management system based on deep learning-based unmanned aerial vehicle swarms and sonic suppression for airport safety, characterized by comprising: a depth camera equipped in each unmanned aerial vehicle (UAV) forming a swarm, having a real-time environment recognition function and accurate object detection and obstacle avoidance functions; a sonic deterrent that repels birds using high-frequency sonic suppression ultrasonic frequencies in the range of 20 to 25 kHz when birds are detected by the depth camera; a monitoring unit equipped in a control tower base station to continuously monitor the swarm of unmanned aerial vehicles (UAVs) equipped with the depth camera; and a deep learning unit integrating Faster R-CNN (Region-based Convolutional Neural Network) and YOLO (You Only Look Once) for real-time bird detection through a system via the monitoring unit.

Description

Technology Field

[0001] The present invention relates to an advanced bird and wildlife management system through deep learning-based unmanned aerial vehicle swarming and sonic suppression for airport safety. More specifically, it relates to an advanced bird and wildlife management system through deep learning-based unmanned aerial vehicle swarming and sonic suppression for airport safety that maintains the safety of aircraft and airports to avoid bird strikes and utilizes unmanned aerial vehicle (UAV) swarming technology equipped with real-time bird detection capabilities using deep learning models and sonic deterrents to solve the critical problems of bird detection and wildlife management near airports by utilizing unmanned aerial vehicle (UAV) swarming and ultimately enhances the safety of both aircraft and bird species. Background Technology

[0003] Generally, in order for an aircraft to land safely on a runway, it must fly along a designated approach path while maintaining a constant glide angle along the center axis of the runway, as shown in Fig. 6; facilities that assist in this are called aircraft landing support facilities. Aircraft landing support facilities include the Instrument Landing System (ILS), Ground Based Augmentation System (GBAS), and Microwave Landing System (MLS).

[0004] The ground equipment of each of these systems is installed at the airport and radiates specific signals to areas such as runways and taxiways. The Glide Path System, mounted on aircraft approaching the airport, receives these signals and generates landing guidance information to guide the runway approach and landing. The generated landing guidance information is provided to the pilot via aircraft instrument panels, such as the Course Deviation Indicator (CDI) or Primary Flight Display (PFD), and the pilot refers to this information to perform a safe runway landing flight.

[0005] Objects present in areas such as runways and taxiways can cause aircraft collisions and resulting casualties, potentially inflicting significant losses on aircraft companies. Some airports utilize radar-based intrusion detection systems. In radar systems, microwave signals are typically transmitted over the runway, and signals reflected from other external objects are detected and analyzed. Higher resolution can be achieved by using radar sensors with smaller wavelengths and higher pulse repetition frequencies.

[0006] However, conventional radar-based intrusion detection systems are hindered from accurately determining whether a moving ground object has intruded into the airport's radio protection zone due to radar clutter video caused by facilities, trees, and grass. Consequently, these systems pre-designate areas such as runways, taxiways, or aprons—where clutter does not occur—as zones for extracting intruder information and monitor for object intrusions within these designated areas. In other words, conventional radar-based intrusion detection systems cannot detect object intrusions across the entire airport.

[0007] In particular, bird strikes pose a serious safety problem in aviation because they occur during critical flight phases. Bird strikes can cause engine damage, windshield obstruction, and structural damage, posing a risk to passenger safety. While aviation authorities worldwide utilize wildlife management programs and containment methods such as noise, visual aids, physical barriers, and chemicals to mitigate bird strikes, challenges remain. Prior art literature

[0009] Published Patent No. 10-2023-0066286 Published Patent No. 10-2024-0088283 The problem to be solved

[0010] Accordingly, the present invention was devised to eliminate the aforementioned problems and utilizes unmanned aerial vehicle (UAV) swarm technology equipped with real-time bird detection capabilities using deep learning models and sonic deterrents to maintain the safety of aircraft and airports in order to avoid bird strikes. It focuses on an advanced bird and wildlife management system through deep learning-based UAV swarms and sonic suppression for airport safety, which solves the critical problems of bird detection and wildlife management near airports by utilizing UAV swarms and ultimately enhances the safety of both aircraft and bird species, and has been completed as a technical task. means of solving the problem

[0012] The present invention, for achieving the above technical objectives, provides an advanced bird and wildlife management system based on deep learning through unmanned aerial vehicle swarms and sound suppression for airport safety, characterized by comprising: a depth camera equipped in each unmanned aerial vehicle (UAV) forming a swarm and equipped with a real-time environment recognition function to enable accurate object detection and obstacle avoidance functions; a sonic deterrent that repels birds using high-frequency sound suppression ultrasonic frequencies in the range of 20 to 25 kHz when birds are detected by the depth camera; a monitoring unit equipped in a control tower base station to continuously monitor the swarm of unmanned aerial vehicles (UAVs) equipped with the depth camera; and a deep learning unit integrating Faster R-CNN (Region-based Convolutional Neural Network) and YOLO (You Only Look Once) for real-time bird detection through a system via the monitoring unit.

[0013] In addition, the present invention is characterized in that the deep learning unit captures aerial images of birds using an unmanned aerial vehicle (UAV), labels each bird by defining a 40×40 pixel bounding box, generates multiple sub-images from the original image using image preprocessing techniques including cropping and augmentation on the image, and trains a deep learning model using the preprocessed images to extract features from hidden layers during the training process.

[0014] In addition, the present invention is characterized in that the Region Proposal Network (RPN) in the Faster R-CNN generates region proposals by predicting anchor box coordinates (xa, ya, wa, ha) and object scores using convolutional features, creates a fixed-size feature map (Pi) in the corresponding region through region of interest (ROI) pooling, the classification head predicts class probabilities using SoftMax activation, and the regression head refines bounding box coordinates (Δx′, Δy′, Δw′, Δh′) to improve localization accuracy. Effects of the invention

[0016] According to the present invention described above, it maintains the safety of aircraft and airports to avoid bird strikes and utilizes unmanned aerial vehicle (UAV) swarm technology equipped with real-time bird detection capabilities using deep learning models and sonic deterrents. By utilizing UAV swarms, it solves the critical problems of bird detection and wildlife management near airports and ultimately enhances the safety of both aircraft and bird species. Brief explanation of the drawing

[0018] FIG. 1 is an exemplary diagram of an advanced bird wildlife management system through deep learning-based unmanned aerial vehicle swarming and sound suppression for airport safety according to the present invention. FIGS. 2 to 7 are exemplary embodiments of the present invention. Specific details for implementing the invention

[0019] The specific details for implementing the present invention will be explained in more detail below with reference to the attached drawings.

[0020] The present invention relates to an advanced bird and wildlife management system through deep learning-based unmanned aerial vehicle swarming and sonic suppression for airport safety. More specifically, it relates to an advanced bird and wildlife management system through deep learning-based unmanned aerial vehicle swarming and sonic suppression for airport safety that maintains the safety of aircraft and airports to avoid bird strikes and utilizes unmanned aerial vehicle (UAV) swarming technology equipped with real-time bird detection capabilities using deep learning models and sonic deterrents to solve the critical problems of bird detection and wildlife management near airports by utilizing unmanned aerial vehicle (UAV) swarming and ultimately enhances the safety of both aircraft and bird species.

[0021] The present invention consists of a depth camera, a sonic deterrent, a monitoring unit, and a deep learning unit.

[0022] By equipping the unmanned aerial vehicle (UAV) swarm presented in this invention with state-of-the-art depth cameras, the system obtains advanced real-time environmental awareness capabilities, enabling accurate object detection and efficient obstacle avoidance. These depth cameras provide essential depth sensing and spatial awareness, allowing the UAVs to accurately identify birds while autonomously navigating complex environments.

[0023] And, when birds are detected by the depth camera, a sonic deterrent is provided to repel the birds using a high-frequency sound suppression ultrasonic frequency in the range of 20 to 25 kHz.

[0024] The suppression system using the above-mentioned sonic deterrent effectively scares away birds by using ultrasonic frequencies. When birds are successfully detected, the onboard foil membrane transducer emits ultrasonic frequencies in the range of 20 to 25 kHz.

[0025] This frequency range is specially selected based on the auditory sensitivity of birds, causing aversion to them but harmless to humans. These small, lightweight foil membrane transducers, similar to those used in Polaroid ultrasonic range modules, generate the desired frequency with sufficient intensity to disperse birds in the area.

[0026] Integrated with the control mechanism of the aforementioned unmanned aerial vehicle (UAV) swarm, this system enables automatic activation and precise deployment upon bird detection, and this sophisticated approach utilizes state-of-the-art technology to mitigate the risk of bird strikes near airports, thereby enhancing aviation safety and enabling wildlife conservation activities.

[0027] In addition, a monitoring unit is provided at the control tower base station to continuously monitor the swarm of unmanned aerial vehicles (UAVs) equipped with the depth camera mentioned above.

[0028] The number of the aforementioned unmanned aerial vehicle (UAV) swarms depends on several other factors, including the size of the airport and the complexity of the airport layout, general bird activity patterns, weather conditions, and specific areas requiring continuous monitoring, such as runways, taxiways, and surrounding airspace.

[0029] In addition, a deep learning unit integrating Faster R-CNN (Region-based Convolutional Neural Network) and YOLO (You Only Look Once) is provided for real-time bird detection through the system via the monitoring unit mentioned above.

[0030] These algorithms excel at recognizing and locating objects within image or video frames, enabling unmanned aerial vehicles (UAVs) to quickly and accurately identify birds around airports. Furthermore, this comprehensive approach, which combines the UAV swarm with advanced sensing and deep learning algorithms, can increase efficiency and effectiveness in mitigating bird-related risks. This not only enhances aviation safety by reducing bird strikes but also enables wildlife conservation activities through humane bird management techniques, rather than conventional methods.

[0031] At this time, the deep learning unit is composed of four steps: capturing aerial images of birds using an unmanned aerial vehicle (UAV), labeling each bird by defining a 40×40 pixel bounding box, generating multiple sub-images from the original image using image preprocessing techniques including cropping and augmentation on the image, and training a deep learning model using the preprocessed images to extract features from hidden layers during the trained process.

[0032] In addition, in the above Faster R-CNN, the Region Proposal Network (RPN) generates region proposals by predicting anchor box coordinates (xa, ya, wa, ha) and object scores using convolutional features, and creates a fixed-size feature map (Pi) in the corresponding region through region of interest (ROI) pooling, the classification head predicts class probabilities using SoftMax activation, and the regression head refines bounding box coordinates (Δx′, Δy′, Δw′, Δh′) to improve localization accuracy.

[0033] Figure 4 shows the faster R-CNN results for successful detection of wild birds, and Figure 5 shows the YOLO results for wild bird detection.

[0034] The above Faster R-CNN (Region-based Convolutional Neural Network) and YOLO (You Only Look Once) models were trained and evaluated with IOU thresholds of 0.3 and 0.5, respectively, which is well illustrated in Fig. 5, showing the precision-return visualization of the Faster R-CNN and R-FCN models using (a) IOU = 0.3 and (b) IOU = 0.5, and Fig. 6, showing the precision-return visualization of the Faster YOLO model using (a) IOU = 0.3 and (b) IOU = 0.5.

[0035] According to the advanced bird and wildlife management system for airport safety through deep learning-based unmanned aerial vehicle swarming and sonic suppression of the present invention as described above, it is possible to maintain the safety of aircraft and airports to avoid bird strikes, and by utilizing unmanned aerial vehicle (UAV) swarm technology equipped with real-time bird detection capabilities using deep learning models and sonic deterrents, it is possible to solve the critical problems of bird detection and wildlife management near airports by utilizing flocks of unmanned aerial vehicles (UAVs) and ultimately enhance the safety of both aircraft and bird species.

[0036] The present invention described above has been explained with reference to an exemplary embodiment illustrated in the drawings, but this is merely illustrative, and it should be made clear to those skilled in the art that various modifications and equivalent alternative embodiments are possible therefrom. Accordingly, the true technical scope of protection of the present invention should be interpreted by the appended claims, and all technical ideas within an equivalent scope should be interpreted as being included within the scope of rights of the present invention.

Claims

Claim 1 An advanced bird and wildlife management system based on deep learning for airport safety using unmanned aerial vehicle swarms and sonic suppression, characterized by comprising: a depth camera equipped on each unmanned aerial vehicle (UAV) forming a swarm, equipped with a real-time environment recognition function, and having accurate object detection and obstacle avoidance functions; a sonic deterrent that repels birds using high-frequency sonic suppression ultrasonic frequencies in the 20–25 kHz range when birds are detected by the depth camera; a monitoring unit equipped at a control tower base station to continuously monitor the swarm of unmanned aerial vehicles (UAVs) equipped with the depth camera; and a deep learning unit integrating Faster R-CNN (Region-based Convolutional Neural Network) and YOLO (You Only Look Once) for real-time bird detection through a system via the monitoring unit. Claim 2 An advanced bird wildlife management system through deep learning-based unmanned aerial vehicle swarming and acoustic suppression for airport safety, wherein, in claim 1, the deep learning unit captures aerial images of birds using an unmanned aerial vehicle (UAV), labels each bird by defining a 40×40 pixel bounding box, generates multiple sub-images from the original image using image preprocessing techniques including cropping and augmentation on the image, and trains a deep learning model using the preprocessed images to extract features from hidden layers during the trained process. Claim 3 An advanced bird wildlife management system based on deep learning for airport safety through unmanned aerial vehicle swarming and acoustic suppression, characterized in that, in the second paragraph, the Region Proposal Network (RPN) in the Faster R-CNN generates region proposals by predicting anchor box coordinates (xa, ya, wa, ha) and object scores using convolutional features, creates a fixed-size feature map (Pi) in the corresponding region through region of interest (ROI) pooling, the classification head predicts class probabilities using SoftMax activation, and the regression head refines bounding box coordinates (Δx′, Δy′, Δw′, Δh′) to improve localization accuracy.

Images