With the development of manufacturing, people benefit from using multiple mobile robots to complete tasks in various scenarios, and multi-robot planning coordination has become a hot research topic in recent decades. ...
详细信息
Mobile UI understanding is important for enabling various interaction tasks such as UI automation and accessibility. Previous mobile UI modeling often depends on the view hierarchy information of a screen, which direc...
详细信息
Supernumerary robotic limb (SRL) is an intelligent platform for providing assistance to humans at extreme conditions in natural human-robot collaboration (HRC) tasks. However, it is still difficult to control the SRL ...
Supernumerary robotic limb (SRL) is an intelligent platform for providing assistance to humans at extreme conditions in natural human-robot collaboration (HRC) tasks. However, it is still difficult to control the SRL by using a simple way and to eliminate the cognitive burden placed on the human operator. of the human operator. Learning that the burden on human operators can be reduced by using Learning from Demonstrations (LfD) which learns repetitive teleoperation tasks. Our study introduces a method for teaching SRL that utilizes Dynamic Time Warping in combination with Gaussian Mixture Model and Gaussian Mixture Regression. The proposed mechanism utilizes a DTW algorithm to align and compare motion patterns between the human teacher and the SRL. Then, the GMM algorithm was applied to encode the motion patterns into statistical models. Additionally, these models are used to reate a task model that is applicable in a broad range of situations through the GMR method, which can be utilized to regulate the motion of SRL in real-time. The proposed mechanism is tested in experiments, and the results show that the SRL can accurately replicate the motion patterns with an average error of less than 10% for the HRC tasks. The proposed teaching mechanism has potential applications in fields such as rehabilitation and industrial automation, where the use of SRL can strengthen human capabilities and improve task efficiency.
In 3C (Computer, Communication, and Consumer Electronics) flexible assembly tasks, due to the frequent updates and the increasing demand for intelligence, traditional automated assembly methods struggle to meet the re...
详细信息
ISBN:
(数字)9798331509644
ISBN:
(纸本)9798331509651
In 3C (Computer, Communication, and Consumer Electronics) flexible assembly tasks, due to the frequent updates and the increasing demand for intelligence, traditional automated assembly methods struggle to meet the requirements for high precision and adaptability. To address these challenges, this paper proposes an assembly method that combines visual reasoning localization with deep reinforcement learning. The method improves the reward function by integrating visual reasoning results with reinforcement learning and trains the model in a virtual environment. Through sim-to-real transfer, the assembly strategies learned in the virtual environment are applied to real-world scenarios, significantly enhancing the assembly accuracy and success rate in real-world settings. Validation on a 3C front camera assembly task demonstrates an assembly accuracy of 0.15 mm and a success rate of 97%.
In the context of the steady development of modern power systems, short-term load forecasting is crucial to the planning and operation of the power grid. Traditional load forecasting methods are effective for linear p...
详细信息
A patient would contract surface muscles as a reaction called muscle guarding when experiencing discomfort and pain during physical palpation. This reaction carries important information about an affected location. Tr...
详细信息
ISBN:
(纸本)9781728190778
A patient would contract surface muscles as a reaction called muscle guarding when experiencing discomfort and pain during physical palpation. This reaction carries important information about an affected location. Training physicians to regulate palpation forces to elicit just enough muscle tension is a challenge using real patients. Tunable stiffness mechanisms enabled by soft robotics can be effectively integrated into medical simulator designs for effective clinical education. In this paper, we propose a controllable stiffness muscle layer to simulate guarding for abdominal palpation training. Designs with soft, fine, and rigid granular jamming, stretchable and non-stretchable layer jamming mechanisms were tested and evaluated as methods to create controllable stiffness muscle. User studies have been carried out on 10 naive participants to differentiate the tense and relaxed abdomen with the proposed jamming mechanisms. Muscle samples made of ground coffee (fine granular jamming) and latex layers (stretchable layer jamming) show good usability in simulating abdomen with different stiffness with at least 75% of the user data exhibits more than 70% of decision accuracy for both tested palpation gestures (single finger and multiple fingers) after short pre-training.
Detection of concrete cracks plays a vital role in structural defect analysis. Classification using machine learning techniques is one of the ways to perform the process. This paper discusses the performance analysis ...
详细信息
ISBN:
(数字)9798350307566
ISBN:
(纸本)9798350307573
Detection of concrete cracks plays a vital role in structural defect analysis. Classification using machine learning techniques is one of the ways to perform the process. This paper discusses the performance analysis of light-weight models used to detect concrete cracks. The Concrete Crack Images for Classification (CCIC) dataset is used in the training and testing phases of NASNetMobile., MobileNet V2., DenseNet121., and VGG-19. Modifications in the networks include layers for binary classification of the data. The quantitative analysis proves the ability of NASNetMobile as a light-weight model while MobileNet V2 can be opted if training time is the constraint. The experimental results validate a trade-off between the usage of light-weight models., especially., NASNetMobile as against regular networks. To prove the tradeoff., the performance of light-weight models is compared with VGG-19 which is a regular model. Analysis proves to waive off the usage of regular models in comparison with the NASNetMobile model.
Real-time performance and accuracy are pivotal for evaluating robotic grasping detection models. To enhance the detection precision while preserving real-time capabilities, this paper presents an EfficientGrasp-gdut n...
详细信息
ISBN:
(数字)9798350364194
ISBN:
(纸本)9798350364200
Real-time performance and accuracy are pivotal for evaluating robotic grasping detection models. To enhance the detection precision while preserving real-time capabilities, this paper presents an EfficientGrasp-gdut network to identify the pixel-level robot grasping positions directly from RGB images. It incorporates one multi-scale convolutional attention (MSCA) module and one multi-scale feature fusion (MSFF) network to improve the robustness in unstructured environments based on EfficientNetV2. The experimental results show that it can attain up to 99.43% accuracy in grasp detection on both images and objects based on the Cornell dataset, with rate up to 26 frames per second (FPS). It has a significant advancement over some leading networks, such as AFFGA-Net and SE-ResUNet.
Underwater image enhancement (UIE) is vital for high-level vision-related underwater tasks. Although learning-based UIE methods have made remarkable achievements in recent years, it's still challenging for them to...
Underwater image enhancement (UIE) is vital for high-level vision-related underwater tasks. Although learning-based UIE methods have made remarkable achievements in recent years, it's still challenging for them to consistently deal with various underwater conditions, which could be caused by: 1) the use of the simplified atmospheric image formation model in UIE may result in severe errors; 2) the network trained solely with synthetic images might have difficulty in generalizing well to real underwater images. In this work, we, for the first time, propose a framework SyreaNet for UIE that integrates both synthetic and real data under the guidance of the revised underwater image formation model and novel domain adaptation (DA) strategies. First, an underwater image synthesis module based on the revised model is proposed. Then, a physically guided disentangled network is designed to predict the clear images by combining both synthetic and real underwater images. The intra- and inter-domain gaps are abridged by fully exchanging the domain knowledge. Extensive experiments demonstrate the superiority of our framework over other state-of-the-art (SOTA) learning-based UIE methods qualitatively and quantitatively. The code and dataset are publicly available at https://***/RockWenJJ/***.
Power Electronic Transformer (PET) is a new type of power electronic device with flexible topology access, which has a wide range of application prospects in AC and DC microgrids. For the control of the MMC-PET system...
详细信息
暂无评论