Sarosh Patel

Infrared small-UAV detection remains difficult because the target often appears as a weak thermal point rather than a clear object. This problem is clear in the SIDD dataset, where most test targets are smaller than 32 × 32 pixels. To address this case, this paper proposes S-Drone-YOLO, a compact YOLO-based detector that maintains a high-resolution P2 prediction path and leverages it carefully during classification. The model starts from a lightweight YOLOv5-style detector. It adds a stride-4 P2 path and replaces the C3 neck blocks with C2fAttn to improve feature reuse before prediction. Two components are then added to the Architecture II design. The Coordinate-Aware Residual C2f Block, CAR-C2f, strengthens the P2 branch using coordinate attention and residual scaling. The P2-Guided Quality-Aware Detection Head (P2-QADH) combines local P2 details with nearby P3 context. It produces a quality map that adjusts the classification logits. The regression branch, output tensor format, and training loss interface remain unchanged. On the SIDD infrared drone dataset, S-Drone-YOLO reaches 0.988 precision, 0.939 recall, 0.699 mAP50-95, and 0.962 F1-score. It uses 6.45 M parameters and 31.3 GFLOPs. Compared with the Architecture I model, recall increases by 0.8 percentage points and mAP50-95 increases by 0.4 percentage points. At the same time, the parameter count decreases by 20.3%, and GFLOPs decrease by 43.7%. Fine-tuning on five RGB UAV datasets and a second thermal dataset (ThermalUAV2UAV) yields F1 scores ranging from 0.941 to 0.999, with an mAP50-95 of 0.843 on the thermal dataset. The background analysis also shows stable F1-scores across sky, sea, city, and mountain scenes. These results suggest that controlled P2 guidance can improve infrared small-UAV detection while keeping the model size practical.

Small-object detection (SOD) remains an important and growing challenge in computer vision and is the backbone of many applications, including autonomous vehicles, aerial surveillance, medical imaging, and industrial quality control. Small objects, in pixels, lose discriminative features during deep neural network processing, making them difficult to disentangle from background noise and other artifacts. This survey presents a comprehensive and systematic review of the SOD advancements between 2023 and 2025, a period marked by the maturation of transformer-based architectures and a return to efficient, realistic deployment. We applied the PRISMA methodology for this work, yielding 112 seminal works in the field to ensure the robustness of our foundation for this study. We present a critical taxonomy of the developments since 2023, arranged in five categories: (1) multiscale feature learning; (2) transformer-based architectures; (3) context-aware methods; (4) data augmentation enhancements; and (5) advancements to mainstream detectors (e.g., YOLO). Third, we describe and analyze the evolving SOD-centered datasets and benchmarks and establish the importance of evaluating models fairly. Fourth, we contribute a comparative assessment of state-of-the-art models, evaluating not only accuracy (e.g., the average precision for small objects (AP_S)) but also important efficiency (FPS, latency, parameters, GFLOPS) metrics across standardized hardware platforms, including edge devices. We further use data-driven case studies in the remote sensing, manufacturing, and healthcare domains to create a bridge between academic benchmarks and real-world performance. Finally, we summarize practical guidance for practitioners, the model selection decision matrix, scenario-based playbooks, and the deployment checklist. The goal of this work is to help synthesize the recent progress, identify the primary limitations in SOD, and open research directions, including the potential future role of generative AI and foundational models, to address the long-standing data and feature representation challenges that have limited SOD.

The Transformer architecture has been the foundational cornerstone of the recent AI revolution, serving as the backbone of Large Language Models, which have demonstrated impressive language understanding and reasoning capabilities. When pretrained on large amounts of data, Transformers have also shown to be highly effective in image classification via the advent of the Vision Transformer. However, they still lag in vision application performance compared to Convolutional Neural Networks (CNNs), which offer translational invariance, whereas Transformers lack inductive bias. Further, the Transformer relies on the attention mechanism, which despite increasing the receptive field, makes it computationally inefficient due to its quadratic time complexity. In this paper, we enhance the Transformer architecture, focusing on its above two shortcomings. We propose two efficient Vision Transformer architectures that significantly reduce the computational complexity without sacrificing classification performance. Our first enhanced architecture is the EEViT-PAR, which combines features from two recently proposed designs of PerceiverAR and CaiT. This enhancement leads to our second architecture, EEViT-IP, which provides implicit windowing capabilities akin to the SWIN Transformer and implicitly improves the inductive bias, while being extremely memory and computationally efficient. We perform detailed experiments on multiple image datasets to show the effectiveness of our architectures. Our best performing EEViT outperforms existing SOTA ViT models in terms of execution efficiency and surpasses or provides competitive classification accuracy on different benchmarks.

This study proposes an approach to 3D modeling of spatial manipulators in the Maple 2023 software environment. Algorithms and program codes have been developed to create computer 3D models of manipulators controlled by generalized coordinates. The implementation of these algorithms and program codes has enabled the creation of three-dimensional computer models of manipulators with clear visual representations of links, their cross-sections, kinematic pairs, grippers, and loads, differing in structure and degrees of freedom while ensuring a comprehensive view from all spatial perspectives. During the motion of the manipulator, complex distributed dynamic loads arise in its links due to their intrinsic masses. These dynamic loads create several challenges: for instance, excessive dynamic loads or significant deformation of the links may lead to failure of the manipulator or a loss of precision in the positioning of the gripper. Such loads significantly impact the design, operation, and reliability of manipulators. The study and understanding of dynamic loads in manipulators are crucial areas in mechanics and robotics, enabling the development of more reliable and efficient systems. The Denavit–Hartenberg method was applied to control the motion of the created computer 3D models of manipulators using generalized coordinates. Using the recursive Newton–Euler equations, the necessary kinematic characteristics of the manipulator’s links were determined for calculating the distributed dynamic loads arising from the intrinsic masses of the links at each cross-section, relative to the local coordinate systems rigidly attached to the links. Algorithms and program codes were developed for controlling the motion of 3D models of manipulators, as well as for constructing visual diagrams of distributed dynamic loads in mutually perpendicular planes, formed by the principal axes of the link cross-sections and the axes passing along the longitudinal axes of the links. The implementation of these algorithms and program codes enabled the generation of distribution diagrams of all dynamic loads in all links of the moving manipulator. These diagrams visually illustrate the changes in direction and magnitude of the distributed dynamic loads in all cross-sections of the links throughout the full cycle of the manipulator’s operation. This allows for the consideration of the identified dynamic loads in the strength and stiffness calculations of the manipulator links, which is essential for the design of new innovative manipulators.

Common challenges in the area of robotics include issues such as sensor modeling, dynamic operating environments, and limited on-broad computational resources. To improve decision making, robots need a dependable framework to facilitate communication between different modules and the optimal action for real-world applications. The Robotics Operating System (ROS) and Reinforcement Learning (RL) are two promising approaches that help accomplish precise control, seamless integration of sensors-actuators, and exhibit learned behavior. The ROS enables seamless communication between heterogeneous components, while RL focuses on learning optimal behaviors through trial-and-error scenarios. Combining the ROS and RL offers superior decision making, improved perception, enhanced automation, and reliability. This work focuses on investigating ROS-based RL applications across various domains, aiming to enhance understanding through comprehensive discussion, analysis, and summarization. We base our evaluation on the application area, type of RL algorithm used, and degree of ROS–RL integration. Additionally, we provide summary of seminal works that define the current state of the art, along with GitHub repositories and resources for research purposes. Based on the review of successfully implemented projects, we make recommendations highlighting the advantages and limitations of RL techniques for specific applications and environments. The ultimate goal of this work is to advance the robotics field by providing a comprehensive overview of the recent important works that incorporate both the ROS and RL, thereby improving the adaptability of these emerging techniques.

Many authors proposed various criteria and concepts to assess manipulability of the robotic system. In this article an isotropy criterion is applied for investigation and synthesis of turning mechanism of a walking robot. Isotropy conditions were derived for the walking robot with orthogonal-type propellers and on its basis the optimal configurations of the robot in terms of force and motion transmission are defined. Solutions of the isotropy equations were found and on the basis of the analysis of isotropic solutions, optimal metric parameters of the robot were found. An experimental prototype of the legged robot is developed.

Interferential Current Therapy is a strong and reliable technique for electrical stimulation or massage. It is based on two current sources with the same strength but different frequencies. But if the current sources are not equal, it can have a devastating effect on the person receiving the massage. Our laboratory results use IFC machines that are easily purchased. Our results on models of human resistance patterns have shown that, if one source’s current is significantly diminished while the second source remains constant, the amount of current that is dumped into the patient will produce heat and burning both exterior and interior. In real life, we have consulted on many accident cases where our lab results are correct. Third degree burns of the skin are common. Also there is burning and scarring of internal organs. A common feature of these accidents is the burning of the skin at one electrode, which causes its current source to drop to almost zero current, since burnt skin has a very large resistance. When this happens, the second source continues to dump current into the patient and cause unwanted heating of skin and organs.

The generalized autoregressive conditional heteroscedastic model (GARCH) is used to estimate volatility for Nifty Index futures on day trades. The purpose is to find out if a contemporaneous or causal relation exists between volatility volume and open interest for Nifty Index futures traded on the National Stock Exchange of India, and the extent and direction of these relationships. A complete absence of bidirectional causality in any particular instance depicts noise trading and empirical analysis according to this study establishes that volume has a stronger impact on volatility compared to open interest. Furthermore, the impulse originating from volatility of volume and open interest is low.

This work presents an Image-Based Visual Servoing (IBVS) technique for an autonomous drone to track and follow an object in 3D space using Image-Based Visual Servoing. Unlike implementing tracking mechanism using GPS positions, Radars or other high-level position determining devices, we present our approach in which we use an onboard camera that captures real-time video stream to processes information such that a given target in 3D space is tracked. We perform IBVS during the flight. A technique to use the video stream and locate the target is presented in this paper. Here we also discuss the mathematical model and system architecture of this technique on an autonomous drone. The advantage of this approach is that the tracking by drones can be made completely autonomous even in GPS-denied environments. We then walk through the applications and future scope of this methodology.