An effort is made to implement the phenomenon called the Aerodynamic Pressure Thrust (APT) for the purpose of effective propulsion of underwater vehicle. The two-dimensional Goldschmied body with boundary-layer ingest...
详细信息
An effort is made to implement the phenomenon called the Aerodynamic Pressure Thrust (APT) for the purpose of effective propulsion of underwater vehicle. The two-dimensional Goldschmied body with boundary-layer ingestion near the stern section is considered for this purpose. This particular shape, which up to this point is being considered mainly for aerodynamic propulsion, can be preferred for autonomous underwater vehicles over conventional streamlined bodies for their large volume-to-length ratio. The wind-tunnel experiments of 1960's by Fabio R. Goldschmied paved a way for development of energy efficient propulsion through proper interaction of aerodynamic design and engine power. Decades later, this eventually led to the development of several futuristic crafts which exploit the Pressure Thrust technique. Similar idea is attempted here to obtain improved pressure recovery behind the aft of a blunt underwater vehicle to generate additional thrust. In the present study, a simplified version of the Goldschmied geometry is considered with a single slot for suction between the fore-body and the stern. The computations are carried out utilizing commercial CFD solver Ansys Fluent. The axi-symmetric shape of the Goldschmied body is employed to generate a fully-structured mesh throughout the entire domain. The simulations are carried out for a range of Reynolds number and the suction pressure. The distribution of pressure at different radial locations is examined and plotted, similar to the original work by Goldschmied, to illustrate the alternate zones of drag and thrust produced along the radius. A sizeable portion of the thrust region in the plot is required to overcome the drag force and to achieve self-propulsion solely by means of boundary layer suction. The results of the computational study indicate a trend in that direction.
Spiking neural network models that have studied how oscillations are generated by recurrent cortical circuits and how they encode information have been focused on describing the encoding of information about external ...
详细信息
Gesture-based robot control offers intuitive interaction between humans and robots, with applications ranging from industrial automation to assistive robotics. However, existing solutions face challenges in achieving ...
详细信息
ISBN:
(数字)9798350362367
ISBN:
(纸本)9798350362374
Gesture-based robot control offers intuitive interaction between humans and robots, with applications ranging from industrial automation to assistive robotics. However, existing solutions face challenges in achieving real-time requirements while ensuring accurate gesture recognition. This paper presents a new edge computing-based approach for real-time control of robots using Electrical Impedance Tomography (EIT) measurements to classify hand gesture numbers in American Sign Language (ASL). Existing solutions for gesture recognition struggle to achieve real-time performance while maintaining accuracy and energy efficiency. This challenge becomes higher in the case of EIT because of its relative complexity. We focus therefore on leveraging the capabilities of the edge device to implement effectively the Convolutional Neural Network (CNN) acceleration. The proposed solution combines hardware-aware optimization techniques to achieve fast and accurate gesture recognition by enabling rapid inference while minimizing energy consumption on a low-power resource-constrained device with Tiny Machine Learning (TinyML) capabilities. The lightweight CNN model required only 10.2 s to train using the Keras library of TensorFlow and achieved an accuracy of 89.37% for 10 sign language classes, with only 66 μs taken to run inference on the hardware-accelerated microcontroller-based device.
Microscopic medical image segmentation is essential for detailed cellular-level analysis in computational pathology, offering insights into disease mechanisms. However, this task is particularly challenging due to the...
详细信息
ISBN:
(数字)9798350363203
ISBN:
(纸本)9798350363210
Microscopic medical image segmentation is essential for detailed cellular-level analysis in computational pathology, offering insights into disease mechanisms. However, this task is particularly challenging due to the high similarity between different cellular classes and variability within each class, especially in densely packed or overlapping cell regions. To this end, we present LEECAU-Net (Latent Entropy-quantized and Efficient Channel Attention-aided U-Net), a U-Net-based model designed to improve segmentation accuracy through a dual-attention mechanism. Built on an Inceptionv4 backbone, LEECAU-Net's encoder integrates Latent Entropy-quantized Channel Attention (LECA) and Efficient Channel Attention (ECA) modules. The ECA module enhances the network's ability to focus on essential channel dependencies, while LECA leverages entropy quantification to prioritize channels rich in information, leading to better differentiation between similar structures. We have tested LEECAU-Net on three challenging publicly available datasets: Triple Negative Breast Cancer (TNBC), Multi-organ Nuclei Segmentation (MoNuSeg), and Cervical Nucleus Segmentation (CNS). Across these datasets, our model achieves higher performance metrics than numerous state-of-the-art approaches reported in the literature. The improved results underscore the effectiveness of LEECAU-Net's architecture in handling the complexity of microscopic medical images, making it a valuable tool for advancing automated cellular analysis in biomedical research. The source code and more results are made available at https://***/Cmatermedicalimageanalysis/LEECAU-Net.
In this paper, we propose a self-organized learning model that can generate behaviors for successfully performing various tasks, unlike conventional systems that could perform only a limited task in a limited environm...
详细信息
In this paper, we propose a self-organized learning model that can generate behaviors for successfully performing various tasks, unlike conventional systems that could perform only a limited task in a limited environment. The model memorizes various relationships between changes in sensory information and a motor command through learning. After the learning, the model can perform various tasks by generating the various behaviors automatically. We investigated the performance of the model by applying it to a mobile robot simulation. In the simulation, we applied two different tasks to the model, which performed the same learning. The results indicate that suitable behaviors for all the tasks emerged spontaneously.
In this paper, we present our first experimental results on the real-time stability analysis of the human operator performing on time-delayed teleoperation system. Due to the fact that human operator behavior is non l...
详细信息
In this paper, we present our first experimental results on the real-time stability analysis of the human operator performing on time-delayed teleoperation system. Due to the fact that human operator behavior is non linear, time varying and the complexity of the techniques to analyze in real-time this kind of systems, we used a discrete linearization of the model and determined for each sample if the system is stable or unstable. In this way, the stability analysis is based on the real-time identified model of the human operator and the Routh-Hurwitz stability criteria is applied to the model. The reason to use Routh-Hurwits instead of Jury criteria is due to the required amount of computational steps. Having determined if the system is stable or unstable in the sampling time, we are able to evaluate the human operator performance. In this way, the proportion of stable and unstable points obtained in the execution of a certain task, provides a performance indicator. The main result of this analysis indicates that human performance, measured as the proportion of stable/unstable points, is increased with repetitive executions of the task and independent of the magnitude of the delay in the visual feedback information.
暂无评论