Autonomous Target Recognition Research Articles

Underwater Side-Scan Sonar Target Detection: YOLOv7 Model Combined with Attention Mechanism and Scaling Factor

Side-scan sonar plays a crucial role in underwater exploration, and the autonomous detection of side-scan sonar images is vital for detecting unknown underwater environments. However, due to the complexity of the underwater environment, the presence of a few highlighted areas on the targets, blurred feature details, and difficulty in collecting data from side-scan sonar, achieving high-precision autonomous target recognition in side-scan sonar images is challenging. This article addresses this problem by improving the You Only Look Once v7 (YOLOv7) model to achieve high-precision object detection in side-scan sonar images. Firstly, given that side-scan sonar images contain large areas of irrelevant information, this paper introduces the Swin-Transformer for dynamic attention and global modeling, which enhances the model’s focus on the target regions. Secondly, the Convolutional Block Attention Module (CBAM) is utilized to further improve feature representation and enhance the neural network model’s accuracy. Lastly, to address the uncertainty of geometric features in side-scan sonar target features, this paper innovatively incorporates a feature scaling factor into the YOLOv7 model. The experiment initially verified the necessity of attention mechanisms in the public dataset. Subsequently, experiments on our side-scan sonar (SSS) image dataset show that the improved YOLOv7 model has 87.9% and 49.23% in its average accuracy (mAP0.5) and (mAP0.5:0.95), respectively. These results are 9.28% and 8.41% higher than the YOLOv7 model. The improved YOLOv7 algorithm proposed in this paper has great potential for object detection and the recognition of side-scan sonar images.

Uncrewed undersea vehicle embedded autonomous target recognition in synthetic aperture sonar imagery

The opacity of the ocean to electromagnetic signals makes communication with Uncrewed Undersea Vehicles (UUVs) extremely difficult, especially for situations where the vehicle is operating far from its center of command and control, and makes mid-operation data exfiltration for external processing untenable. This leaves operator intervention and control limited necessitating the need for advanced and adaptive autonomy that can create situational awareness from mission data provided by vehicle sensors. Long range high-resolution imagery generated by synthetic aperture sonar (SAS) is a high fidelity data product that can be further processed by both machine learning techniques and traditional image processing techniques to create computer aided detection and classification (CAD/CAC) capabilities to feed that developing autonomy need. This work examines two separate detection and classification pipelines that have been used in parallel to create contact sets that are actionable for onboard decision processes.

Sonar acoustic feature interpretation using braid manifolds

Autonomous target recognition has been a long-standing problem with complementary techniques based on classic subspace-based methods and recent machine cognition techniques proposed across several decades. Despite some success in both approaches, robust detection and meaningful interpretation of target features stay an open challenge. The primary bottleneck is the entanglement of multi-dimensional spectral features, some of which are intrinsic to target composition and geometry, and others originate from environmental reverberation and background clutter. This challenge is further compounded by unpredictable non-linear overlap between target features, which may themselves morph as a function of experimental conditions such as ping angle, range, frequency, and other factors. Furthermore, there is a compelling need to represent target features in a compact representation that lends itself to meaningful physical interpretation by a domain expert as well as robust interface with popular machine learning architectures. In this light, we will provide a broad overview of our recent and continuing work with braid manifolds, a morphological construct based on topological continuity of target features, that bridges this gap between physical model-based target representation and autonomous machine-learnt representation of target features. Representative results based on experimental field data on different target sizes and experimental conditions will be presented.

3D Object Recognition and Localization with a Dense LiDAR Scanner

Dense scanning is an effective solution for refined geometrical modeling applications. The previous studies in dense environment modeling mostly focused on data acquisition techniques without emphasizing autonomous target recognition and accurate 3D localization. Therefore, they lacked the capability to output semantic information in the scenes. This article aims to make complementation in this aspect. The critical problems we solved are mainly in two aspects: (1) system calibration to ensure detail-fidelity for the 3D objects with fine structures, (2) fast outlier exclusion to improve 3D boxing accuracy. A lightweight fuzzy neural network is proposed to remove most background outliers, which was proven in experiments to be effective for various objects in different situations. With precise and clean data ensured by the two abovementioned techniques, our system can extract target objects from the original point clouds, and more importantly, accurately estimate their center locations and orientations.

Feature-based autonomous target recognition and grasping of industrial robots

Feature-based autonomous target recognition and grasping of industrial robots

Design of Airport Obstacle-Free Zone Monitoring UAV System Based on Computer Vision.

In recent years, a rising number of incidents between Unmanned Aerial Vehicles (UAVs) and planes have been reported at airports and airfields. A design scheme for an airport obstacle-free zone monitoring UAV system based on computer vision is proposed. The system integrates the functions of identification, tracking, and expelling and is mainly used for low-cost control of balloon airborne objects and small aircrafts. First, a quadcopter dynamic model and 2-Degrees of Freedom (2-DOF) Pan/Tilt/Zoom (PTZ) model are analyzed, and an attitude back-stepping controller based on disturbance compensation is designed. Second, a low and slow small-target self-identification and tracking technology is constructed against a complex environment. Based on the You Only Look Once (YOLO) and Kernel Correlation Filter (KCF) algorithms, an autonomous target recognition and high-speed tracking plan with great robustness and high reliability is designed. Third, a PTZ controller and automatic aiming strategy based on Anti-Windup Proportional Integral Derivative (PID) algorithm is designed, and a simplified, automatic-aiming expelling device, the environmentally friendly gel ball blaster, which features high speed and high accuracy, is built. The feasibility and stability of the project can be verified through prototype experiments.

Development of autonomous target recognition and scanning technology for pulse-echo ultrasonic propagation imager

This article proposes a pulse-echo ultrasonic propagation imaging system that is capable of autonomous target recognition and scanning an area of a customizable shape for the non-destructive evaluation of structural defects. The proposed system employs through-the-thickness bulk waves for ultrasonic inspection, which is achieved by joining two laser beams: one each for the ultrasonic wave generation and sensing. Moreover, the system is capable of autonomously suggesting a suitable inspection area for a specimen placed in front of the scanning head. The scan area delimitation algorithm uses the specimen image and ascertains the specimen border by means of edge and contour detection operations, following which a scan area closely conforming to the specimen boundary is suggested. The system can scan an area of any arbitrary shape, thereby preventing any wasteful operations that may result from fixed shape (rectangular or square) scanning. A Q-switched laser is used for generating the aforementioned ultrasonic waves, while a laser Doppler vibrometer is used for sensing these signals. A dual-axis automated translation stage is applied for raster scanning of the specimen, and a data acquisition card is employed for taking measurements. A camera mounted on the scan head is used for imaging the specimen for the scan area detection. Graphical user interface software controls all the individual blocks of the system, while implementing the required image processing, scan area detection, signal acquisition, signal processing, and result display. The graphical user interface is created in C++ using the Qt framework. Moreover, Qt Widgets for Technical Applications is used for the result display, and the Open Source Computer Vision Library is employed for the implementation of basic image processing algorithms. Multi-threading is used for real-time updating of the scan results while the scanning is ongoing.

Computer Simulation of AUV Navigation System Using Fish Lateral Line Sensing Mechanism

Based on the study of fish lateral line sensing mechanism, integrated fluid dynamics, boundary layer theory, coupling theory and so on. To establish a neural network model similar to the fish neuromas model. After the numerical calculation and simulation analysis to simulate fish lateral line system, we can get the conclusions to apply to autonomous underwater vehicle (AUV) navigation and target recognition. Provided a new method for AUV navigation and environment perception. Computer simulation results presented that based on the fish lateral line sensing mechanism to build a perception model, AUV can identify a new water condition distinguishably. This discovery can be used in navigation and positioning of AUV.

Modeling for a System Operating Characteristic for Wide Area Mines

Traditional landmines are limited in their attack range and information providing ability. There are currently several wide area search mines in the research and development phase to overcome these disadvantages. Progress on the individual technologies is promising, but there are insufficient analytical tools for evaluating the effectiveness of these concept mines. This paper examines some of the modeling aspects of wide area search mines with Autonomous Target Recognition (ATR) capability. The optimal employment of autonomous wide area search mines is addressed. The specific scenarios considered involve several mines searching a battle fields for targets in the presence of false targets. All relevant parameters are extracted from intelligence information, the sensor performance specification, and the vehicle performance specification. Analytic system effectiveness measures are derived using applied probability theory. Optimal schedules for controlling sensor threshold during a mission are derived. These schedules establish a system operating characteristic. An increase in system effectiveness is demonstrated when parameters are dynamically controlled during a mission.

Optimization of an Autonomous Weapon System's Operating Characteristic

Optimal employment of autonomous wide area search munitions or unmanned aerial vehicles is considered. The air vehicle searches a battle space for stationary targets in the presence of false targets and/or clutter. Control problems are formulated to achieve optimal scheduling of the control variables P TR, the probability of target report, which is a reflection of the sensor's threshold setting, and Q, the area coverage rate. These optimal control problem formulations yield solutions that maximize the probability of target attack, and constraints to also satisfy hard bounds on the probability of false target attacks may also be applied. Analysis has shown that the optimal solutions are sensitive to the particular relationship between the autonomous target recognition module's receiver operating characteristic parameter, c , and area coverage rate, Q . In this paper the relationship is developed from first principles for both passive and active sensors, and the optimal P TR and Q schedules are obtained.

Model-based target recognition using laser radar imagery

Autonomous target recognition can be assisted by using CO 2 laser radar data. Of these data, range data provide 3-D geometric information and Doppler data detect boundaries of moving targets. A powerful 3-D feature extraction algorithm based on the Hough transform is used to obtain the orientations and dimensions of the target. This information is then utilized by an inference procedure that recognizes targets based on the available evidence from the sensory data. The experimental results using actual laser radar imagery are successful and the procedure can be used for the future development of a model-based expert system for target recognition.