Human motion prediction based on the biological signal has been improved because of the improvement of an artificial intelligent (AI) technology. Conventional methods to predict a human motion have been based on a machine learning such as a neural network, a regression model and so on. From the above, the model by them cannot generate the model whose calculation process is not clear. Then, it is impossible to interpret the relationship between input and output information. In this paper, the element description method (EDM) was applied to generate the model and the prediction of human motion and the analysis about muscle characteristics had been done. An EDM is one of the system identification methods and it is possible to interpret the relationship between input and output because it can generate a block diagram of the model. Especially, the fingertip force and the surface-electromyography (sEMG) of the muscles which move the joint were focused in this paper Moreover, the fingertip force was estimated from the SEMG and it was also done to analyze the muscle characteristics based on the model by an EDM.
The accurate knowledge of the synchronous reluctance machine (SyR) model is essential for advanced control implementation and high-fidelity simulation. The accuracy of experimental constant-speed identification methods is mainly influenced by inverter nonlinearities; however, factors such as the digital control effect, the current harmonics effect of the machine under the test, and the speed harmonics effect of the prime mover should be considered. Therefore, this paper investigates the influence of the aforementioned effects on the accuracy of model identification for SyR motor drives.
Batteries are widely applied in industrial systems, promoting industrial production efficiency. The system prognosis needs to establish a battery health monitoring model and maintain regular maintenance, improving reliability and safety while reducing unnecessary economic losses. An accurate data-driven state-of-health (SOH) estimation method is crucial, as it mainly relies on the information related to battery SOH values in the extracted features. Therefore, we propose a statistical-based feature extraction method, which makes it physically meaningful and effective. To explore more information, we propose a feature analysis framework to analyze battery features, including visualization, correlation, and importance analysis. We utilize a random forest regression model as an effective learner and validate it based on NASA’s real-world dataset. The feature analysis contents are beneficial for understanding the role and importance of features in SOH estimation tasks, which can also provide an extended scenario for feature extraction of batteries.
Realizing both dexterity in a large workspace and mechanically efficient architecture is a challenge for robotic manipulators. Conventional industrial robots, including serial and parallel robots, have suffered from the trade-off between the weight and the size of the workspace. On the other hand, cable-driven parallel robots (CDPRs) perform excellent lightness of the platform in a large workspace. However, its workspace is surrounded by a box, which leads to a large occupied space. This paper presents a hybrid cable-driven parallel robot with an articulated manipulator. The end effector is a platform whose orientation is directly controlled by four cables. The proposed architecture realizes 6-DOF motion by a light body. Since the tendons do not go through the intermediate joints of the articulated manipulator, the mechanical efficiency of the cables is improved compared to conventional tendon-driven robots. The forward kinematics is analyzed, and the numerical solution using the Newton-Raphson method is derived. The position control method, considering the statics of the platform, is explained. The experiments of position control confirm the performance of both translational and orientational motion.
Recent research on quadruped robots has been achieving high-performance motion control based on optimization and reinforcement learning. However, there is still ongoing research aimed at demonstrating high-performance motion based on simple and dominant dynamic principles. In this paper, we proposed a novel control approach that projects Spring-Loaded Inverted Pendulum (SLIP) dynamics to articulated legs, utilizing admittance control based force observer within a rotating workspace (RWFOB). Unlike other legged robots that depend on sensor-based estimation of external forces, the proposed method presents an alternative approach that reduces the reliance on sensors. Additionally, we introduce a comprehensive control framework for quadruped robot motion control, establishing the connection between trunk and SLIP-realized leg movements using Jacobian. The effectiveness of the proposed framework as a robust and reliable trunk feedback controller is validated through simulation and experiments.
This study presents a comparative analysis of stray flux, rotor current and stator current signals to identify field winding faults in synchronous motors, employing an advanced combination of high-order spectral analysis and time-frequency wavelet techniques. The work proposes a time-frequency analysis framework using Wavelet transforms in the bispectral domain, which is applied under various operating regimes of the motor (under starting and steady-state). The results of this research are promising and demonstrate the potential for seamless integration of these diagnostic techniques into electrical machine condition monitoring systems. This approach improves the accuracy of fault detection in synchronous motors and can contribute to the improvement of maintenance strategies in autonomous systems in industrial applications.
The current study proposes a network control structure for small low-cost drones like the Parrot Mambo mini-drone. The structure is composed of an inner loop running on the drone, and an outer loop running on a remote computer. The inner loop controls the attitude and altitude of the drone based on Kalman filter estimations from the onboard sensors. The outer loop ensures position tracking based on measurements from OptiTrack cameras. A time delay compensator is added to address the constraints imposed by wireless network communications between the drone and the remote computer. Experimental results using Parrot Mambo drones show good stability and tracking performances, despite model uncertainty and time delay.
The electrification of heating and mobility sectors has spurred a surge in power demand, necessitating the integration of Demand Response (DR) and Demand-Side Management (DSM) to mitigate potential grid congestion. This paper presents the data-efficient Internet of Things Constrained Application Protocol (CoAP) as alternative to IEC 61850 for the communication and data exchange of an actively managed renewable energy community (REC). The paper outlines the architectural framework for typical use cases, providing insights into control strategies for Photovoltaic (PV) generation, energy storage systems (ESS), and Electric Vehicle (EV) charging stations (CS). The research presented highlights the significant communication overhead associated with IEC 61850 and a leveraging CoAP machine-to-machine (M2M) mapping based on data measurements. Further, the necessary data traffic for active management of an exemplary REC is extrapolated and data savings by application of a CoAP mapping illustrated.
Induction motors are electric machines widely used in mechanisms requiring rotary motion. Despite being robust machines, they are prone to failure. Among the most common and challenging failures to differentiate are those caused by mechanical issues such as imbalance and misalignment. Although various methodologies for diagnosing mechanical faults in these motors exist, most employ computationally intensive techniques or require costly equipment. This study proposes a methodology for automatically diagnosing imbalance and misalignment faults using a low-cost infrared sensor. This involves processing the sensor data by calculating two statistical indicators (mean and kurtosis) from regions of interest and employing a classifier based on the GradientBoost algorithm. The methodology is validated on an induction motor with variable load using a DC motor.
The widespread implementation of electric vehicles (EVs) depends crucially on the optimization of charging infrastructures. In this article the application of a dual active bridge (DAB) multi-module topology for an electric vehicle (EV) DC charger is presented. The proposed DC-DC converter allows for working with a wide range of DC output voltages at nominal power. The PISO (Parallel Input Series Output) / PIPO (Parallel Input Parallel Output) configuration have been studied and analyzed to achieve maximum performance. An algorithm is proposed to determine the optimal configuration at each operating point, ensuring efficient and reliable charging across different battery voltages, boosting the adaptability of the charging system. The algorithm is validated using MATLAB/Simulink and PLECS, using GaN MOSFETs for better performance. Furthermore, the simulation software is employed to calculate the system losses, and the results are analyzed to assess the charger’s performance.
Object detection remains a pivotal challenge in computer vision, with the YOLO series establishing itself as the industry standard for efficient solutions. Concurrently, the inverted bottleneck convolution module has been instrumental in the success of the MobileNet family, enabling efficient feature extraction. In this research, we integrate the inverted bottleneck convolution module into the state-of-the-art YOLOv8 object detection model, strategically incorporating it into the low-resolution regions of the network. The proposed model achieves superior performance on the COCO dataset, outperforming the original YOLOv8 by 1% in mAP50:95 score while employing fewer parameters and requiring fewer Giga FloatingPoint Operations per Second (GFLOPS), contributing to a more lightweight and efficient architecture. Remarkably, the proposed model exhibits comparable inference speeds on various hardware platforms, ensuring real-time performance and practical deployment capabilities. This balance between accuracy and efficiency positions our model as a compelling alternative to YOLOv8, offering superior object detection performance while maintaining resource-efficient operations. Our work highlights the potential for cross-pollination of architectural innovations across different families of convolutional neural networks, paving the way for further exploration and optimization of efficient object detection models.
Maize leaf diseases are critical determinants of maize yield and quality. With the decrease of population and arable land resources, deep learning-based machine vision methods provide obvious advantages in recognizing and detecting maize leaf diseases in precision and cost, and it has become a hotspot of current research. However, there are many types of maize leaf diseases, over 90 types of maize leaf diseases can be found in the worldwide. The diagnosis of one disease or several diseases cannot fulfill the needs of practical applications. In addition, deep learning model training requires a large amount of datasets and label, and it is very difficult to obtain comprehensive dataset samples and labels with existing supervised learning methods. A method that relies on limited data samples and label is desperately needed. In this paper, we propose a novel method for detecting maize leaf diseases using semi-supervised learning techniques, which incorporates a soft teacher model into the Faster R-CNN network framework. By leveraging a small set of labeled data, the model generates pseudo-labels for a large pool of unlabeled data. This innovative strategy significantly enhances the detection performance of the model, resulting in more accurate and reliable disease detection outcomes. Results of the experiment in the PlantVillage public dataset indicated that the method achieved a superior detection result under different labeling percentages of data. In particular, its performance reaches 57.2% when the labeled data is 50%, which is a 25.1% improvement compared to the supervised learning method.
This paper proposes a vision-based framework for a 7-degree-of-freedom robotic manipulator, with the primary objective of facilitating its capacity to acquire information from human hand demonstrations for the execution of dexterous pick-and-place tasks. Most existing works only focus on the position demonstration without considering the orientations. In this paper, by employing a single depth camera, MediaPipe is applied to generate the three-dimensional coordinates of a human hand, thereby comprehensively recording the hand's motion, encompassing the trajectory of the wrist, orientation of the hand, and the grasp motion. A mean filter is applied during data pre-processing to smooth the raw data. The demonstration is designed to pick up an object at a specific angle, navigate around obstacles in its path and subsequently, deposit it within a sloped container. The robotic system demonstrates its learning capabilities, facilitated by the implementation of Dynamic Movement Primitives, enabling the assimilation of user actions into its trajectories with different start and end poi
TOPSAR has operational advantages over conventional synthetic aperture radars (SARs) by steering the beam in the azimuth direction. However, because of its complex observation geometry, azimuthal degradation occurs when the illumination mode is highly-squinted. In the context of highly-squinted TOPSAR (HS-TOPSAR), targets symmetrically positioned along the flight path exhibit nearly identical Doppler histories, leading to ambiguities, whereas beam steering narrows and skews the spectrum, diminishing signal orthogonality and introducing spectral folding. Consequently, the Doppler gradient diminishes to a degree that makes precise measurement challenging, thereby constraining fine azimuth angular resolution during imaging. To this end, a hybrid-domain algorithm utilizing azimuth iterative focusing is proposed to address azimuth-variant characteristics, resolving ambiguities and enhancing spectral orthogonality for image reconstruction, as verified by simulation results and metrics.
In recent years, the convergence of the Industrial Internet and edge computing accelerates the evolution of edge computing towards edge intelligence. The new architecture of Industrial Internet and edge computing requires that industrial edge applications can handle hard real-time production tasks while satisfying the high-reliability demand of industrial sites. Traditional industrial software development cannot cope with such demands. In this paper, the computational design model for contract-based design is applied in automatic code generation for industrial edge applications to solve the above problems. The proposed method is mainly for iteration to improve the generation process from requirement to actual code. The intermediate model generated by the computational model is verified with the wind turbine generator system, a typical application of industrial edge computing systems. The paper provides an efficient and flexible solution for rapidly reconfiguring and optimizing the intermediate model in response to changing requirements, which contributes to the automatic code generation for industrial edge applications. Moreover, this approach can meet diverse system performance and maximize resource utilization to reduce costs significantly.
This paper presents a method for real-time and accurate identification of plant diseases using the state-of-the-art YOLOv8 (You Only Look Once Version 8) object detection model. Addressing issues of inaccurate detection and unstable accuracy rates encountered in traditional disease diagnosis processes, particularly concerning small-sized disease targets, we introduce an approach that incorporates an attention mechanism for detecting small disease lesions. The experimental dataset comprises annotated image samples of various common plant diseases affecting a range of crops including legumes, strawberries, and tomatoes. After preprocessing and transforming the dataset into a format suitable for input into the YOLOv8 model, and conducting targeted optimization training on the model, we compared its accuracy with commonly used disease detection models. Our identification model demonstrated relatively high efficiency in recognizing diseases in these three types of crops, thus carrying significant research implications for intelligent agricultural monitoring.
This paper concerns the event-triggered finite-time course control problem of an unmanned surface vehicle (USV) under the network environment. To facilitate investigation, the nonlinear dynamic of the course control system is firstly modeled by a class of state-dependent uncertain systems. Then a output feedback framework of course control is established for the USV subject to wave-induced disturbance, where an event generator with dynamic triggering scheme and a logarithmic quantizer are introduced to save communication resources. By means of slack variable technique, a novel co-design criterion of output feedback controller and event-triggered scheme is developed to guarantee the course control system finite-time bounded with a desired H-infinity performance. Finally, the effectiveness of the proposed method is verified through a practical simulation case.
A constantly growing demand for high-precision components for special applications requires ever more precise and flexible machine tools. As a result, industrial robots are often used. Despite their versatility, their precision and accuracy may not be sufficient for certain tasks. For example, high accuracy is required for the manufacturing of free form optics or ultra precision mirrors. Even small errors in the optical surface can cause an unacceptable imaging error. Various techniques to improve this are already state of the Art. Tracking the robot’s end effector with an external measuring device is frequently used to calibrate or correct the robot during its operation. The Multi-Aperture Positioning System (MAPS) is one of the measuring devices capable of doing so. With MAPS, the position and orientation of multiple coordinate systems can be measured simultaneously. The position of an end effector can be tracked online by attaching a light source-based target to it. The orientation, on the other hand, could previously only be calculated when the machine was stationary. The positions of three LEDs are required for this. The challenge is to measure the three positions and find the correspondence between each measurement and the respective light source. The current solution to this problem is to modulate the LEDs in time. These are switched on and measured one after the other. If the machine is in motion, this would lead to a spatial offset between the measurements. If, on the other hand, they are all switched on at the same time, the relationship between the light source and the measured value is lost. This is because the spots in the image can no longer be distinguished from each other and overshadow the markers. This article presents a new approach to ensure correspondence between the light source and the measured value when the LEDs are switched on at simultaneously. In this method, the light sources are labeled with a shape that is transferred to the image. An algorithm for separating the spot sets and a newly developed LED target are presented. This method enables MAPS to measure the end effector of machines such as robots in 6DOF while they are moving.
In recent years, the adoption of reproducible, easily deployable container technology has grown across a variety of sectors. Containers ensure that each service runs in an independent environment because they provide a consistent execution environment, are easy to scale, and allow for efficient versioning and deployment. In a Software Defined Vehicle (SDV), each service must not interfere with the others and must be able to be updated independently. Due to these challenges, ease of scalability, efficient versioning and deployment are important research topics. In this paper, we analyze the applicability of container technology to ROS2-based automotive architectures for autonomous driving systems.
Indoor lighting design is to combine natural light sources and artificial light sources in a specific space and makes complete planning to meet the needs of users for lighting, to achieve the effect of creating a certain scenario. Due to the mainstream of LED lamps and the popularity of the Internet of Things, the lighting design market is growing. Compared with traditional lighting equipment, smart lighting can bring users a more comfortable lighting experience. It can be turned on and off through a computer or mobile device, or the lamps can be controlled and dimmed according to the user’s preference, making the lighting environment more ergonomic and reducing the power consumption of the lighting system. In this paper, an indoor automatic dimming system based on machine learning and particle swarm optimization (PSO) algorithm is implemented. The experiment is carried out remotely. Through neural network and particle swarm optimization algorithm, the dimming commands of lamps that best fit the current environment are found to meet the illumination requirements of the users. This study focuses on the operation and application of machine learning and PSO algorithm, and verifies the correctness and feasibility of the proposed automatic dimming system by comparing the simulation and measured results with the traditional daylight responsive dimming system (DRDS), and verifies the improvement of dimming accuracy and energy saving performance of the proposed method.
