Explore the vast range of titles available.

The advent of automation in transportation technology plays a crucial role in revolutionizing modern logistics systems. This research focuses on the design, modeling, and performance analysis of a LuggAge Carrying BoT (LABO) that combines kinematic modeling and statistical data analysis to predict path trajectories. The LABO system incorporates an advanced adjoint transformation model, distinct from conventional Point of Entry formula models, to establish precise path calibration and compensate for load-induced deviations. In autonomous mode, a BOX-Ljung statistical model predicts path trajectories by analyzing speed, payload, and positional data. Experimentally, LABO demonstrated optimal performance, carrying a payload of up to 14 kg at a maximum speed of 0.24 m/s on rough surfaces. The motor driver configuration was tested under varying loads, showing a decline in speed from 0.3 m/s at 2 kg payload to 0.25 m/s at 12 kg payload. Kinematic analysis was performed using a serial configuration with a 20-degree-of-freedom model, yielding a 0.27% improvement in path precision when compared to Denavit–Hartenberg models. The proposed LABO system offers a reliable, user-friendly, and cost-effective solution for transporting heavy luggage, significantly reducing the physical burden on elderly or disabled individuals, while the BOX-Ljung model ensures precise path prediction for improved autonomous navigation.

Inspired by the UNet architecture of semantic image segmentation, we propose a lightweight UNet using depthwise separable convolutions (DSUNet) for end-to-end learning of lane detection and path prediction (PP) in autonomous driving. We also design and integrate a PP algorithm with convolutional neural network (CNN) to form a simulation model (CNN-PP) that can be used to assess CNN’s performance qualitatively, quantitatively, and dynamically in a host agent car driving along with other agents all in a real-time autonomous manner. DSUNet is 5.12 × lighter in model size and 1.61 × faster in inference than UNet. DSUNet-PP outperforms UNet-PP in mean average errors of predicted curvature and lateral offset for path planning in dynamic simulation. DSUNet-PP outperforms a modified UNet in lateral error, which is tested in a real car on real road. These results show that DSUNet is efficient and effective for lane detection and path prediction in autonomous driving.

The current network risk assessment model often ignores the impact of attack cost and intrusion intention on network security. In order to better solve the problem of information security defense strategy selection and accurately assess the target network risk, this paper proposes an attack path prediction method based on game model.The atomic attack probability is calculated by vulnerability value, attack cost and attack benefit. The static risk assessment model is established combined with Bayesian belief network quantitative attack graph. And the dynamic update model of intrusion intention is used to realize the effective prediction of attack action under rational assumption, which provides the basis for dynamic defense measures of attack surface. The experimental results verify the feasibility and effectiveness of the model and method.

Mobile robots face significant challenges in obstacle detection and safe navigation in populated indoor environments. While commonly used, traditional sensors such as a depth camera and a 2D LiDAR exhibit limitations in handling dynamic objects and complex scenarios. Additionally, most current perception pipelines rely on discrete sensors and rarely integrate motion prediction or object tracking, leaving planners blind to rapidly changing obstacles. To address these issues, this study introduces an Obstacle Detection and Tracking Framework that integrates state-of-the-art object detection algorithms—including YOLOv11, MobileNet-SSD and EfficientDet-D0—with multi-object tracking using both DeepSORT and ByteTrack. This tracking information is proactively incorporated into the robot’s occupancy grid through the occupancy grid generation module, where regions predicted to be occupied by moving obstacles are dynamically assigned higher costs. Detected and tracked objects are projected onto a 2D bird’s eye view using depth data, providing spatial and temporal awareness of the environment. To enhance path planning, we add a sensor-fused trajectory predictor. For each tracked object, RGB-D and 2D LiDAR are fused via a Kalman filter to estimate position, velocity of objects for path prediction. This framework bridges the gap between perception and planning systems and provides actionable insights for selecting suitable algorithmic configurations in real-world robot deployment scenarios.

Various studies have shown that the ant colony optimization (ACO) algorithm has a good performance in approximating complex combinatorial optimization problems such as traveling salesman problem (TSP) for real-world applications. However, disadvantages such as long running time and easy stagnation still restrict its further wide application in many fields. In this study, a saltatory evolution ant colony optimization (SEACO) algorithm is proposed to increase the optimization speed. Different from the past research, this study innovatively starts from the perspective of near-optimal path identification and refines the domain knowledge of near-optimal path identification by quantitative analysis model using the pheromone matrix evolution data of the traditional ACO algorithm. Based on the domain knowledge, a near-optimal path prediction model is built to predict the evolutionary trend of the path pheromone matrix so as to fundamentally save the running time. Extensive experiment results on a traveling salesman problem library (TSPLIB) database demonstrate that the solution quality of the SEACO algorithm is better than that of the ACO algorithm, and it is more suitable for large-scale data sets within the specified time window. This means it can provide a promising direction to deal with the problem about slow optimization speed and low accuracy of the ACO algorithm.

Due to the high cost of specially customised presses and dies and the advance of machine learning technology, there is some emerging research attempting free-form sheet metal stamping processes which use several common tools to produce products of various shapes. However, tool path planning strategies for the free forming process, such as reinforcement learning technique, derived from previous path planning experience are not generalisable for an arbitrary new sheet metal workpiece. Thus, in this paper, a generalisable tool path planning strategy is proposed for the first time to realise the tool path prediction for an arbitrary sheet metal part in 2-D space with no metal forming knowledge in prior, through deep reinforcement (implemented with 2 heuristics) and supervised learning technologies. Conferred by deep learning, the tool path planning process is corroborated to have self-learning characteristics. This method has been instantiated and verified by a successful application to a case study, of which the workpiece shape deformed by the predicted tool path has been compared with its target shape. The proposed method significantly improves the generalisation of tool path planning of free-form sheet metal stamping process, compared to strategies using pure reinforcement learning technologies. The successful instantiation of this method also implies the potential of the development of intelligent free-form sheet metal stamping process.

This paper explored the application of fully connected neural networks and transfer learning techniques in laser forming, explicitly focusing on predicting scanning paths in multi-stage forming processes. Due to the time-consuming data collection, this research built imaginary data to replace experimental data with training in neural network efficiency. The research developed a machine learning model trained on imaginary datasets that can predict the scanning paths of experimental datasets. However, the test results on experimental data indicated that the imaginary data generation method led to a performance that did not meet expectations. Therefore, this paper also compared the differences between the imaginary and the experimental data. Furthermore, this research adopted transfer learning to address the issues of overfitting and low prediction accuracy encountered by fully connected neural networks at higher forming stages. It explored the fine-tuning and final-layer methods. The results suggest that the fine-tuning method can effectively enhance the model’s generalization capability, achieving better performance in complex forming stages. Compared to the fully connected network, it reduces nearly 35% of training iterations and improves test accuracy by 3–5 mm.

The loading and unloading operations of smart logistic application robots depend largely on their perception system. However, there is a paucity of study on the evaluation of Lidar maps and their SLAM algorithms in complex environment navigation system. In the proposed work, the Lidar information is finetuned using binary occupancy grid approach and implemented Improved Self-Adaptive Learning Particle Swarm Optimization (ISALPSO) algorithm for path prediction. The approach makes use of 2D Lidar mapping to determine the most efficient route for a mobile robot in logistical applications. The Hector SLAM method is used in the Robot Operating System (ROS) platform to implement mobile robot real-time location and map building, which is subsequently transformed into a binary occupancy grid. To show the path navigation findings of the proposed methodologies, a navigational model has been created in the MATLAB 2D virtual environment using 2D Lidar mapping point data. The ISALPSO algorithm adapts its parameters inertia weight, acceleration coefficients, learning coefficients, mutation factor, and swarm size, based on the performance of the generated path. In comparison to the other five PSO variants, the ISALPSO algorithm has a considerably shorter path, a quick convergence rate, and requires less time to compute the distance between the locations of transporting and unloading environments, based on the simulation results that was generated and its validation using a 2D Lidar environment. The efficiency and effectiveness of path planning for mobile robots in logistic applications are validated using Quanser hardware interfaced with 2D Lidar and operated in environment 3 using proposed algorithm for production of optimal path.

Typhoons are one of the most destructive types of disasters. Several statistical models have been designed to predict their paths to reduce damage, casualties, and economic loss. To further increase prediction accuracy, two key challenges are (1) to extract better nonlinear 3D features of typhoons, which is hard due to their complex high-dimensional properties, and (2) to combine suitable 2D and 3D features in a proper way to improve predictions. To address these challenges, this paper presents a novel spatio-temporal deep learning model named Attention-based Multi ConvGRU (AM-ConvGRU). To automatically select high response isobaric planes of typhoons when considering their whole 3D structures, AM-ConvGRU leverages the Residual Channel Attention Block (RCAB). Furthermore, it integrates a novel model named Multi-ConvGRU to extract large-scale nonlinear spatial features of typhoons. Moreover, the approach relies on a Wide & Deep framework to fuse the traditional Generalized Linear Model (GLM) with the proposed AM-ConvGRU model. To evaluate the designed approach, extensive experiments have been conducted using real-world typhoons data from the Western North Pacific (WNP) basin obtained from both the China Meteorological Administration (CMA) dataset and the EAR-Interim dataset maintained by the European Centre for Medium-Range Weather Forecasts (ECMWF). Results show that the proposed method outperforms state-of-the-art deep learning typhoon prediction methods. The source code is available on GitHub with the following link: https://github.com/xuguangning1218/Typhoon_Path .

The advancements in the education sector made e-learning more popular in recent years. The velocity of learning content creation and its availability is also growing exponentially day after day. It is challenging for a learner to find a learning path for a course with a vast content repository. So, recommending a learning path helps the learners streamline the learning materials systematically and achieve their goals. This article proposes a learning path recommendation approach focused on knowledge building and learning performance analysis. The model considers both static and dynamic learner parameters for learning path generation. The difficulty level of the learning resources is tuned based on the real-time performance analysis of the students. The learning resources are recommended based on learning preferences and the ability of a learner to learn the specific learning resource. The model also predicts the learning time and the expected score for each learner. Root Mean Square Deviation and Coefficient of Determination (R-Squared error) measures are used to find the correctness of the prediction. The model is also checked for its adaptivity to the learners’ changing behavior and diversity of the LOs recommended for different learners. Ninety-six undergraduate learners participated in the study. The experimentations are conducted with 530 LOs from selected courses. The comparison results with three existing models show a better performance from the proposed approach with an average accuracy rise of 30% in learning path prediction based on the expected duration of learning 27.8% in expected score prediction with the second-best performing model. It is observed that the real-time learning analytics using the implicit learner log data benefits the recommendation process. LO rating strongly indicated the enhancement of learner satisfaction and experience with a rise of 25.5% when comparing the rating share with the second-best model.