Machine learning
Researchers from Tsinghua University have made significant breakthroughs in autonomous driving technology, leveraging online mapping and crowdsourcing trajectory data to enhance lane segment perception and topology reasoning. By incorporating prior information into an online mapping model, the team's approach has shown to significantly outperform current state-of-the-art methods. The
Machine learning
A Japanese patent filed by the inventors Kanzaki, Jun, Oyama, Yasunori, and Santo, Yoshiki for a robot system and method for predicting the life of a regenerative resistor has been published online on June 3, 2025. The patent, assigned to Kawasaki Jukogyo Kabushiki Kaisha, aims to provide a solution to
Machine learning
A new study published in Discover Applied Sciences has presented a significant advancement in malaria detection and classification using an ensemble of Convolutional Neural Networks (CNN) and Vision Transformers (ViT). The proposed model, designed by researchers at Kyambogo University, has demonstrated superior performance compared to existing methods, achieving an accuracy
Machine learning
Researchers from Huaibei Normal University have proposed a novel residual spatial state-space (RSSS) block to enhance spatial feature extraction in medical image segmentation. The RSSS block combines the Mamba module for capturing global dependencies with a receptive field attention convolution (RFAC) module to extract location-sensitive local patterns. This breakthrough has
Machine learning
A team of researchers from Liaoning Petrochemical University has proposed a new non-end-to-end adaptive graph learning algorithm that addresses the limitations of existing methods in predicting traffic flow. The algorithm incorporates multiple modules to capture complex dependencies and spatial relationships in road networks, resulting in significant improvements in predictive performance.
Machine learning
Researchers from Dong-A University, in collaboration with other institutions, have developed a hybrid machine learning framework that combines an artificial neural network and a genetic algorithm to optimize chemical compositions of shielded metal arc weld metals for achieving targeted mechanical properties. This approach has demonstrated high predictive accuracy and potential
Machine learning
A new study on artificial neural networks has been published, aiming to overcome the limitations of machine learning in estimating wheat traits. Researchers from the University of Natural Resources and Applied Life Science in Tulln, Austria, have developed a simulated dataset using the radiative transfer model (RTM) PROSAIL and inverted
Machine learning
Researchers at Fudan University have introduced Topo-Field, a novel framework for efficient topometric mapping in robotics. Topo-Field leverages a brain-inspired hierarchical layout-object-position (LOP) field model to integrate object and layout information, enabling the construction of a topometric map from a learned neural representation. This approach successfully bridges the gap between
Machine learning
A patent application has been filed for an apparatus that generates motion control data for a robot to facilitate nonverbal communication. The apparatus is designed to modify target motion based on a breathing motion pattern of a second user, simulating a more natural and empathetic interaction. This technology has the
Machine learning
Researchers from Oakland University have made significant breakthroughs in the field of robotics and artificial intelligence, particularly in few-shot learning techniques. According to a study published in Frontiers in Robotics and AI, they have developed a novel framework incorporating a Global Attention Mechanism (GAM) and Latent Dirichlet Allocation (LDA) generated
Machine learning
Researchers from Shandong University of Technology have made significant progress in the field of robotics, proposing a novel control framework for multi-axle active suspension vehicles to traverse impassable abrupt roads. The study, which was financially supported by the Natural Science Foundation of Shandong Province and the National Natural Science Foundation
Machine learning
Scientists at the University of College London (UCL) have made significant strides in robotics by developing a single-layer multimodal sensory skin that can capture a wide range of data over a large surface area. This innovative skin uses a highly sensitive hydrogel membrane to provide a soft interface, emulating the
