This PDF file contains the front matter associated with SPIE Proceedings Volume 13284, including the Title Page, Copyright information, Table of Contents, and Conference Committee information.

With the rapid development of intelligent robots, the application of cooperative robotic arm in dangerous and high-precision situations has become a research hotspot. The trajectory planning of its arm is one of the difficult problems. On the basis of mathematical modeling and demonstration of the cooperative manipulator, in view of the inherent defects and problems of the traditional VPF algorithm, this paper proposes an improved velocity potential field algorithm considering the direction factors and obstacle velocity factors, which can stably avoid dynamic obstacles and ensure the safety of operators and equipment. The paper improves the RRT algorithm to enhance its running speed during the operation of dual robotic arms.

This study presents an improved random forest algorithm for motor short-circuit fault diagnosis. The classification accuracy of the model is significantly improved by data preprocessing, time and frequency domain feature extraction, and parameter optimization. The experimental results show that the improved random forest algorithm performs well in the fault classification task and has obvious advantages over the traditional methods. This study provides an effective solution for motor fault diagnosis and provides a reference for further research in related fields.

To enhance the control effectiveness of hybrid magnetic bearings (HMBs), this paper proposes a PSO-based fuzzy PID control algorithm. The algorithm optimizes the fuzzy controller's output parameters using the Particle Swarm Optimization technique, effectively addressing the limitations caused by fixed parameters and improving control accuracy. The effectiveness of the PSO algorithm optimization is verified through simulations of a hybrid magnetic bearing system built in Simulink. Compared to the fuzzy PID control algorithm, this method demonstrates a faster response and smaller overshoot, achieving better results in controlling the vibration displacement of rotors supported by hybrid magnetic bearings.

In this paper, an improved gray wolf optimization linear active disturbance rejection control (IGWO-LADRC) had been proposed by improving the grey wolf population initialization, convergence factor, and grey wolf position updating strategy of the traditional grey wolf optimization algorithm. The problems of standard gray wolf optimization algorithm in linear active disturbance rejection control parameters tuning, such as low efficiency of optimization search and easy fall into local optimal solution, had been solved. A brushless DC motor control system based on IGWO-LADRC and a brushless DC motor control system based on gray wolf optimization linear active disturbance rejection control (GWO-LADRC) had been built. The comparison had shown that IGWO-LADRC had a fast optimization-seeking speed, high control accuracy, quick response speed, and strong anti-interference ability.

To solve the drive slip prevention problem of distributed drive electric vehicles under extremely complex road conditions, this paper proposes a multi-agent body distributed control method based on modularized architecture. Firstly, the modular approach is used to reconstruct the whole vehicle structure, consider each hub motor wheel as an independent agent, and establish the slip rate model of each wheel agent. Then, a distributed model predictive control method based on multi-agent is designed to solve the drive anti-slip problem considering multi-constraints and multi-objectives. Finally, the anti-slip performance is verified by simulation under low adhesion and split road conditions. The results show that the distributed control strategy under the modularized architecture can synergistically control the wheels to achieve good anti-slip effect. This provides a theoretical and practical reference for further development of modular vehicle in the future.

Steel plate corrosion causes huge economic losses to industrial manufacturing environments and may induce accidents. Therefore, monitoring the corrosion rate of steel plates is very important in the industrial field. However, the accuracy of traditional steel plate corrosion rate monitoring is difficult to meet the requirements of industrial sites. For this reason, it is particularly important to design a high accuracy steel plate corrosion rate monitoring model. In order to solve this problem, this paper proposes a high accuracy steel plate corrosion rate monitoring model based on Dilated convolution, named HDCNN. Starting from the fine-tuning paradigm of Dilated convolution, a new plug-and-play self-attention mechanism module is proposed, named JAM module, to help the model extract more effective information, thereby improving the monitoring accuracy of the model. At the same time, in order to verify the effectiveness of the model, this article self-made a set of data sets with multiple complex working conditions (different lights and different experimental scenarios), and conducted comparative experiments under the data sets with multiple complex working conditions. Experimental results show that among the three groups of mainstream models, the HDCNN model has the highest test accuracy (Accuracy: 93.1785 %).

Crop output can be greatly enhanced by using greenhouses, and efficient environmental management techniques are necessary to achieve this goal. The greenhouse environment, being a multifactor, highly linked, nonlinear complex system, is vulnerable to multiple interferences that might affect the overall stability of the control system. Controlling greenhouse settings in an accurate, efficient, and intelligent way is still a major challenge. This article examines the three developing stages of greenhouse control to assess the common intelligent control methods used nationally and internationally in greenhouses. It focuses on how three important intelligent control techniques—fuzzy control, decoupling control, and neural network control are now being applied. Moreover, it recommends that multi-variable system control and control model optimization be given top priority in future research on greenhouse intelligent control to offer guidance and assistance for more sophisticated and intelligent greenhouse management in the future.

The machinery and equipment manufacturing industry are developing at an unprecedented speed, making the inspection of mechanical parts increasingly complex. In this work, we propose a neural network-based accuracy evaluation method for mechanical parts. This method involves the collection of a large number of machining and inspection data sets of mechanical parts, the removal of incomplete or erroneous data through data cleaning, the extraction of key parameters through the use of feature extraction techniques, and the implementation of standardized processing to ensure data consistency. A multi-layer feedforward neural network (MLP) was constructed, the ReLU activation function was employed to address the nonlinear transformation of the hidden layer, and the Adam optimization algorithm was utilized for training to enhance the convergence speed and stability of the model. The experimental results demonstrate that the proposed method is an effective and reliable approach for predicting the machining accuracy of mechanical parts, and it is well-suited to meet the needs of industrial processing sites.

This paper aims at the problem of lightweight design of a leading tank truck production enterprise in China. It analyzes the methods of lightweight and combines them with the actual requirements. Using ANSYS software, the FEM model of the whole LPG tanker is built. The strength and stiffness information of the tanker are calculated. The paper studies the lightweight of the tanker by using high-strength steel. Then, the stress and stiffness of the lightweight LPG tanker are checked. The fatigue life is calculated and analyzed by HyperWorks. It provides a reliable basis for the design and manufacture of the LPG tanker.

Aiming at the collaborative encirclement and interception problem of multi-agent vehicles with terminal line-of-sight (LOS) angle constraints, this paper designs a cooperative guidance control by using the multi-agent leader-follower structure coordination theory. Regarding the collaborative interception problem between the leader vehicle and the follower vehicle, a multi-agent leader-follower structure consensus protocol is proposed. Based on this consensus protocol, a time-collaborative guidance control is designed to achieve a synchronous attack on the target by the follower vehicles and the leader vehicle. For the terminal LOS angle constraint of each vehicle, the terminal sliding mode control method is combined with a variable exponential power reaching law to design a guidance control that satisfies the constraint. Finally, in this article, the designed time collaborative guidance control with terminal LOS angle constraints is used to conduct simulations to verify its feasibility for high-speed target interception problems.

Low cost and zero emission have always been the goal of general aircraft. In this paper, the life cycle cost estimation model of light electric aircraft is studied, and the aerodynamics, flight performance, weight, and economic analysis model of electric aircraft is established. A multi-disciplinary design optimization method for light electric aircraft based on life cycle cost is presented. Taking a four-seat electric aircraft as an example, the parallel subspace optimization algorithm is used to optimize the model and find the optimal solution. The life cycle cost estimation method of electric aircraft and the multidisciplinary design optimization method of the overall design of electric aircraft studied in this paper can not only be directly applied to the conceptual design stage of electric aircraft but also provide technical support for the future research and development of electric aircraft in China.

This design is based on Arduino microcontroller and OPEN_ The thermal powered car with MV camera. Most of the smart cars on the market are based on a drive model powered by an electric motor and battery, which is lacking in cost and endurance. Therefore, this article improves the original battery and motor drive model to be driven by a Stirling engine, combined with OPEN_ The MV, HC-05 module, and infrared control module, the intelligent control of the car has been achieved. The camera combines image processing technology to achieve the car tracking function, Bluetooth transmission to achieve remote control, and the infrared tube module to achieve obstacle avoidance function. The experimental results show that this design can not only save energy but also accurately achieve functions such as car tracking and obstacle avoidance, which have certain practical value.

In the present era, robot industry is experiencing rapid development. Manipulators have emerged as essential robotic systems that play crucial roles in various domains. Ensuring high precision and reliability of manipulators holds paramount significance, particularly within the industrial sector. Numerous approaches have been explored to enhance the precision and reliability of manipulators by calculating and compensating for errors. However, there are some shortcomings in the existing methods. To address the existing issues and enhance the overall performance,based on the modified DH (MDH) model, this paper proposes an error prediction and compensation approach that combines back propagation (BP) neural network and particle swarm optimization (PSO). In this method, the attitude of IRB1410 manipulator is described by MDH model, and the forward and inverse kinematics are simulated by Matlab software. To complete the training of the BP neural network model, a small disturbance is randomly given to the manipulator within the constraint condition to obtain the corresponding data set. The model serves as a simulator for predicting errors, while the particle swarm optimization algorithm is employed to compute the compensation value, thereby circumventing local optima. Combined with the particle swarm optimization algorithm, the root mean square error before and after compensation is reduced by 78.65%. It can improve the accuracy of the manipulator, so that it can better complete people's needs.

In order to analyze the influence of the formation flight on the aerodynamic characteristics of the bird wings, two rows of upstream and downstream blades are set up to simulate the flying birds in the formation. In this study, DNS method was used to simulate the aerodynamic characteristics of blades.The FieldView 20 software was used for flow visualization and the Tecplot software for detailed data analysis.Through the study of aerodynamic principles, we can further explore the aerodynamic characteristics and causes of formation flight, which is of great significance for the research and design of aircraft or drone formations. By analyzing the aerodynamic performance of blades and its functional characteristics , the influence and mechanism of the formation flight were obtained in this study. The study shows that the tail vortex periodically grows and falls off the upstream blade to produce the vortex street, and the vortex street affects the near surface flow characteristics of downstream blade. Under the influence of the wake of the upstream blade, the change range of the low pressure area around the downstream blade increases, and a unidirectional vortex is produced in the front of the blade. This eddy current increases the area of the low pressure area and destroys the generation of the tail vortex, and increases lift and reduces drag.

This design scheme aims to create an intelligent electronic dimming table lamp with manual and automatic control functions, which can achieve flexible brightness adjustment, automatically turn off the lamp when walking, and provide a more intelligent experience through a real-time display of power and distance. In addition, the design also supports Bluetooth control on mobile phones, allowing users to achieve the same functions as buttons through their phones, which can meet the growing demand for smart furniture. This design scheme not only achieves the basic requirements of turning on, adjusting brightness, and automatically dimming the electronic dimming table lamp, but also provides additional functions such as working time setting and recording, automatic reduction of brightness, real-time display of voltage, current, and power. To achieve the above functions, the design scheme adopts multiple key components, including a 4x4 matrix keyboard for operation control, a 128×64 0.96-inch OLED display screen for displaying various parameters, and a GY-39 light intensity sensor module for detecting ambient light brightness. Using the STC15F2K60S2 minimum system board as the main controller, the output PWM signal controls the bidirectional thyristor voltage regulation module to adjust the voltage, achieving precise control of the brightness of the incandescent lamp. At the same time, the IM1253B AC/DC AC coulometer is used to measure and display the working voltage, current, and power data of the table lamp. The software control adopts a method combining interrupt and polling. It applies ultrasonic modules to remind users to turn off the lights in a timely manner by measuring the distance between people and lamps, thereby achieving energy conservation and environmental protection.

The aim of this study is to investigate the design approach for smart home control panel interfaces (SHCPI) in the context of aging, to enhance the sustainable user experience for elderly users, and to optimize usability. The research methodology involves applying design thinking to identify user needs by diverging creatively along two logical pathways using the Mash-up method, resulting in the construction of a hierarchical model. Weight analysis is conducted using the analytic hierarchy process (AHP) to inform interface design decisions. Subsequently, the fuzzy comprehensive evaluation method (FCE) is employed to assess satisfaction and acceptance of the design by elderly users. Findings reveal that seven design elements with higher weight rankings have been prioritized to inform practical implementation. The comprehensive evaluation indicates a decision value of 86, suggesting that the design scheme adequately meets user requirements. The study concludes that the proposed design enhances user experience for the elderly, offering insights for innovative strategies and directions for aging design in smart home control panels, while also presenting a valuable reference for research on aging interfaces across various domains.

In view of the characteristics of independently controllable torque of each wheel of a distributed drive electric wheel vehicle and rapid adjustment speed, this paper studies the stability control problem during its driving. Using vehicle dynamics theory, the vehicle stability evaluation parameters yaw angular velocity and center of mass slip angle are selected as control variables, and considering their error change rates, a yaw moment sliding mode controller is designed, using coordination to eliminate the coupling influence of the two control variables. The control strategy generates the yaw moment required to maintain the stable driving of the vehicle and effectively distributes it to the four drive wheels. The CarSim-Simulink simulation test platform was used to verify the designed control strategy by taking typical angular step test conditions as an example. The results show that the adopted sliding mode coordinated control strategy can significantly improve the stability and safety of distributed drive electric wheel vehicles.

With the gradual rise of the development trend of informatization and intelligence in industrial manufacturing, an improved NSGA-II algorithm is proposed to study the problems of numerous orders, long completion time, low equipment utilization, and long time to generate scheduling solutions only by MES in the assembly production of highly customized workshops, which constructs an intelligent scheduling model with the minimum completion time and the maximum equipment loading rate. Adopting adaptive crossover and variational probability design allows the algorithm to better adapt to the changes in the solution space during the search process. On the initialization of the next generation population after each iteration, a local search is performed based on the optimized solution set of the previous generation, which improves the algorithm's convergence speed and the optimized solution's accuracy. Adjusting the calculation order of fast nondominated ordering and congestion distance during the iteration process and still calculating fast non-dominated ordering and congestion distance after the iteration is completed, the optimal solution can be screened out and the algorithm's performance can be significantly improved. A simulation experiment of scheduling a complex order of a military product is designed to verify the feasibility and effectiveness of the improved algorithm in intelligent scheduling.

How to develop an optimal emergency dispatching plan in the response phase to minimize casualties and property losses is a hot topic of current research. In recent years, the research results of scholars at home and abroad on intelligent scheduling methods for emergency rescue have been systematically organized and reviewed from the dimensions of scheduling models and solution algorithms. The current research status of the intelligent scheduling model is discussed according to three types: single-objective scheduling model, multi-objective scheduling model, and two-layer planning model, and the solution algorithms are compared and analyzed from the perspectives of improvement scheme and optimization objective, and the development trend of the research on emergency rescue scheduling method is looked forward to.

Electro-hydraulic actuator (EHA) system is a key component of the ship's electro-hydraulic control system, but the internal failure mechanism of EHA is extremely complex, and it is limited by insufficient detection methods in practical systems, making fault diagnosis of EHA a great challenge. To address this issue, this paper proposes a fault diagnosis method based on Continuous Wavelet Transform (CWT), Convolutional Neural Network (CNN) and Transformer. Firstly, the collected one-dimensional time series signal of the sensor is converted into a two-dimensional time-frequency map through CWT. Then, a hybrid module of CNN-Transformer is used to extract local and global fault features, and finally the classification results are output through softmax. Conduct EHA fault simulation experiments to validate the proposed method. The research results indicate that the proposed method has a fault diagnosis accuracy of 96.33%, achieving high-precision diagnosis of EHA faults.

In order to solve the problem of low welding efficiency and unstable welding quality of thin plate non-structural parts, a set of automatic welding machine is developed by using weld seam tracking technology based on laser vision sensor. The mechanical system and control system of the welding machine are designed. By identifying the characteristic points of the weld seam, the position deviation between the welding torch and the weld seam is calculated and fed back to the controller to adjust the position of the welding torch, thus the adaptive and automatic welding of the butt weld of the thin plate is realized. The experimental results show that the automatic welding machine can better track and modify the welding trajectory. It can meet the needs of on-site welding construction of marine construction.

This article provides a detailed analysis of the structure and function of the high-power nuclear power circulation pump gearbox, and explores its failure modes and effects in depth. A detailed description of the structural composition of the gearbox was provided, including the transmission system, lubrication and cooling system, feedback system, etc. The system structure and classification of the vertical planetary reduction gearbox were elaborated.A systematic analysis was conducted on the failure modes of the gearbox using the Failure Mode and Effects Analysis method, and the impact of different failure modes on the system was evaluated.

The squirrel cage motor has many advantages, but its obvious weakness is the small starting torque. When used to drive large inertia loads, the phenomenon of "big horses pulling small cars" often occurs. This article optimizes and improves the rotor groove shape of a regular three-phase induction motor, thereby increasing the starting torque of the motor. At the same time, in response to the problem of increased heat generation and decreased efficiency caused by the improved high slip, permanent magnet plates were added to the motor rotor to improve the power factor and efficiency of the motor by utilizing magnetic pulling force, reducing the thermal load. The simulation results show that the improved motor has achieved a significant increase in starting torque while ensuring efficiency.

Aiming at the environmental control of cable accessories during the installation process and the overall portable processing equipment for the working environment of high-voltage cable accessories, this paper developed a self-sensing processing device for high-voltage cross-linked cable accessory installation environment, which integrates the functions of purification, temperature and humidity. And this device is designed with a wireless sensing real-time monitoring module. It can achieve real-time monitoring of temperature, humidity and cleanliness while providing feedback to the environmental treatment device. Based on the environmental detection results, the device can autonomously control to stop and operate, achieving self-sensing function.

The working principle of the constant speed valve was introduced, and the static characteristics of the constant speed valve were analyzed through the establishment of a mathematical model. The modeling and simulation of the constant speed valve were carried out using AMESim software. The influence of inlet pressure on the dynamic response of constant speed valve is analyzed. And by changing the main spring stiffness of the constant speed valve, the size of the oil outlet, the area of the fixed orifice, the outer diameter of the valve core and other structural parameters, the influence on the performance of the constant speed valve is analyzed. It provides a basis for further improving the dynamic characteristics and design optimization of the constant speed valve.

Bipolar radiofrequency (RF) heats the deeper layers of the skin to denature collagen. This treatment will tighten the skin and reduce the appearance of wrinkles. However, due to the high electrical impedance in the epidermis, RF energy transmission requires the application of a conducting medium to the surface of the skin to increase electrical conductivity. In this study, the finite element analysis is used to investigate the influence of the conductivity of the RF conducting medium on the heating performance of the bipolar RF technique, which has acquired the optimal conductivity of the conduction medium used in household RF beauty instruments. For bipolar RF, the thermal conduction equation of the four-layer skin tissue with a surface conducting medium is firstly derived, and then a finite element model of bipolar RF skin is established by COMSOL Multiphysics to simulate tissue heating with a conducting medium which has a conductivity of 900-2500 μs/cm. The results showed that the highest temperature range in the target area was 41.22-42.23 ℃. Compared to the case without the conducting medium, the temperature increased by approximately 2.4-5.1%, and the heating rate increased by 2.51-10.42%. Taking into account the safety and effectiveness of RF treatment, for a household RF beauty instrument with an output frequency of 1 MHz and a peak-to-peak voltage of 110 V, the conducting medium with an optimal conductivity of 900 μS/cm was more suitable.

Bipolar radiofrequency (RF) cosmetology technology has the advantages of safety and painlessness, long duration, high patient tolerance, easy operation, etc., and has a wide range of applications. Electrode diameters and spacing significantly affect subcutaneous temperature distribution, and a systematic study of this is necessary. In this paper, the effect of different electrode diameters and spacing on the subcutaneous temperature distribution of bipolar radiofrequency aesthetics was investigated based on simulation analysis using COMSOL Multiphysics. The finite element skin model of bipolar radiofrequency was first established, and then the subcutaneous temperature distribution of bipolar radiofrequency with different electrode diameters and spacing was simulated by finite element analysis. The results showed that when the electrode diameters were gradually increased, the location of the hottest temperature target was basically unchanged, but the heating area was enlarged and the temperature rise was significant; On the other hand, when the electrode spacing was increased, the location of the hottest temperature target was gradually deepened, and the heating area was enlarged, but the rate of temperature rise was reduced. In particular, when the electrode diameters are in the range of 4-6 mm, the spacing is in the range of 2.37-4.25 mm, after being heated for 20 s, the depth of the hottest temperature target is about half of the electrode spacing, and the area of the 41-43 °C temperature range is the largest, which shows that it is suitable for household cosmetology device of bipolar RF.

In recent years, there has been unprecedented enthusiasm for live events, and high-quality live events rely heavily on the coordination of stage lighting. This paper presents a method that applies virtual simulation technology to lighting design. The method is based on the Unreal Engine and GrandMA2 OnPC, enabling virtual simulation of lighting design in real scenes, and providing more intuitive and realistic simulation effects for stage performances, film and television productions, and other fields of lighting design. Additionally, to showcase the overall stage effect, this system also incorporates video design and camera position design, improving stage elements and creating a more realistic simulation. Experimental validation demonstrates that this lighting simulation system can accurately simulate real lighting effects, with fast and flexible scene transitions and lighting adjustments, thereby enhancing the efficiency of lighting design.

For multi-axis CNC machining of complex parts, the dynamic prediction of contour error is the basis and guarantee for on-line adjustment of machining parameters and high-quality machining. During the operation of the machine tool, the change of the pose of each feed shaft makes the machine tool show obvious time-varying dynamic characteristics, which makes the contour error difficult to predict. Aiming at this problem, a dynamic prediction method of contour error of multiaxis machine tool driven by mechanism-data hybrid is proposed. Taking the two-axis feed system of a five-axis machining center as the specific object, the dynamic modeling and solution are carried out, and the rigid-flexible coupling transfer function matrix is constructed as the mechanism model. The data set of feed shaft pose and natural frequency is obtained by modal experiment, and the data model is constructed by machine learning algorithm. The time-varying rigid-flexible coupling transfer function matrix is obtained by combining the mechanism model with the data model. Taking it as the controlled object, a time-varying servo feed system is built in MATLAB/Simulink to realize the dynamic prediction and verification of contour error. Finally, the above model is integrated into VE² to develop the digital twin model of the machine tool, so as to realize the dynamic presentation of contour error and tracking error. The research results have reference value for constructing the contour error model under the influence of time-varying dynamic characteristics.

Recent advance in the Microsoft Kinect has provided great opportunities for general intelligent application of motion sensing interaction. The somatosensory technology has demonstrated epoch-making significance in the field of human-computer interaction (HCI), which enriches users’ new interactive experience with machine without any handheld devices. This new generation of natural interaction is making the communication between human and computer more and more accessible since it takes nature as the core and provides speech movement tracking and posture-computer interface. In this paper, we conduct a comprehensive survey on the applications of Kinect-based motion sensing technology in HCI, which aims to further probe the issues of more humanized interaction and promote the deep integration of motion sensing technology and daily application. Moreover, the sustainability of motion sensing was particularly explored, indicating that the technology could be a trigger for promoting smart social activities, favouring intelligent human-machine development, and, in general, improving the quality of life.

As medical and cognitive sciences advance, the proportion of individuals diagnosed with Autism Spectrum Disorder (ASD) has reached 1% of the global population. A significant majority, approximately 90%, of these individuals suffer from sensory overload, a condition stemming from neural processing disorders that profoundly affects their daily lives. The sensory processing differences makes it difficult for the general public to empathize with those with ASD, leading to misconceptions and biases. This study develops a multisensory interactive device designed to simulate the experience of sensory overload for individuals with ASD in urban street environments. This device enables non-ASD individuals to undergo visual, auditory, and tactile sensory overload, addressing the gap in bodily perception that previous works overlooked. By fostering interactive engagement, this research aims to dismantle perceptual and empathetic barriers, thereby enhancing societal understanding and support for individuals with ASD.

The information filtering task is an important human-computer interaction task in the air defense and anti missile command and control system, which has the characteristics of wide content range and multiple adaptation scenarios. Currently, it is mainly completed by humans operating the relevant areas of the display and control panel interface through mouse, keyboard, and other operations. When the information load is large and the number of task scenarios increases, the input operation efficiency is low. Therefore, it is necessary to study command and control interaction methods suitable for mixed reality environments. The existing methods for operation in Hololen2 mainly include gesture action, eye movement, and speech, with different effects. How to reasonably combine and improve the efficiency of information filtering and input interaction in MR environment is an urgent problem that needs to be solved. This article mainly analyzes the human-machine interaction characteristics of MR devices and designs a simulation information filtering operation interface experiment in the MR environment to explore the completion effect of tasks under different interaction matching schemes, in order to judge the advantages and disadvantages of the schemes.

The process of exoskeleton robot design is a complex system requiring the cooperation and integration of machinery, medicine, automation, and computer technology. As one of the high-tech industries, the field of exoskeleton robotics contains a large number of scientific and technological achievements and innovative technologies in its patent literature, and efficiently utilizing the knowledge therein is crucial for improving the efficiency and design quality of exoskeleton robotics. Patent knowledge is cross-disciplinary and multi-disciplinary in nature, leading to difficulties in integrating knowledge, as well as a lack of organization and management methods. These problems have led to underutilization of patent knowledge information resources in this field. In order to make effective use of the information and data resources of patents, shorten the design cycle, and provide knowledge for the designers, according to the operations objectives, an exoskeleton robot technology knowledge graph was established. On the basis of patent analysis, entities and relationships in the knowledge graph were defined, and the knowledge ontology layer was established. By fine-tuning the UIE model to extract the deep innovation content knowledge of the patent and filling it into the data layer.

The efficacy of interface design for manned submersibles operating at full ocean depth extends beyond the subjective impressions of the designers; it is imperative that the layout is user-friendly to address the physiological and psychological needs of submariners engaged in prolonged activities within confined spaces. This directly influences their performance efficiency and the safety of deep-sea exploration tasks. In this study, a Philips BDM4037UW 40-inch curved monitor was used to replicate the cockpit's display and control interface, segmented into seven zones. The E-prime software facilitated the presentation of visual stimuli. Subjects situated at the mock-up interface were presented with visual stimuli and required to respond by keypress. Each zone randomly displayed a single-digit number ranging from 0-9, prompting participants to discern whether the number was odd or even by pressing the "F" and "J" keys accordingly. The display utilized three background color groups—white, black, and gray, and five foreground color groups for the numbers—white, black, red, green, and blue. Behavioral metrics, such as accuracy and reaction time, were logged via the software's backend. Eye-tracking data were also collected, encompassing fixation details (visual intake count, visual intake frequency, visual intake duration total and visual intake duration average), saccade details (saccade count, saccade frequency, saccade duration average and saccade velocity average), and blink details (blink count, blink frequency, blink duration total and blink duration average), serving as evaluative indicators of the interface's human-computer interaction. Findings suggest that visual information processing is more efficient with specific color pairings: a red foreground on white or gray backgrounds and a green foreground on a black background. For optimal usability, it is recommended that key, important, and secondary information be strategically positioned within different areas of the display and control interface for deep-sea operations.

In high-risk operational environments, pilots are susceptible to stress reactions, which can adversely influence their cognitive process and operational performance. Thus, it is crucial to devise effective methods for assessing pilots' stress levels. This research developed a pilot stress classification model employing eye-tracking and heart rate variability (HRV) features within a simulated aviation task context. Utilizing a neural network algorithm, a three-tier stress classification model was established, demonstrating a five-fold cross-validation accuracy of 80.12%. The results provide methodological support for the evaluation of stress states in pilots. Furthermore, the result can provide a foundation for pilots' training and selection processes, thereby improving their capacity to manage stress and ensure aviation safety.

In recent years, lower limb rehabilitation exoskeleton robots have been widely used in patients with lower limb movement disorders caused by neurological injuries and other disorders to provide long-lasting and standardized rehabilitation training assistance, and have therefore attracted the attention of the market and many researchers. Currently, the lower limb rehabilitation exoskeleton technology is relatively mature, but how to design more intelligent and personalized lower limb rehabilitation exoskeleton machines is still a significant problem. Through literature research, this paper compares and analyzes different types of lower limb rehabilitation exoskeletons in terms of their different structural forms, drive modes, and control strategies, to understand their advantages and disadvantages as well as their applicability, and puts forward an outlook on the development of lower limb rehabilitation exoskeletons in the future. Eventually, through the discussion and summary, we conclude that the future of lower limb rehabilitation exoskeleton robots will be developed in the direction of lightweight, safe, intelligent, etc., to provide rehabilitation patients with more efficient and perfect rehabilitation assistance services.

This paper discusses the current status and future prospects of power robot standardization. Power robots, as “special forces” serving the power system, are widely used in power production, operation and maintenance, overhaul and other aspects to ensure the safe and stable operation of the power grid. This article firstly outlines the definition, development history and application status of power robots in the power system, especially China's leading position and achievements in the research and application of power robotics. Subsequently, the article analyzes the important role of power robot standardization in promoting technological innovation, facilitating industrial development, and improving the efficiency of operation and maintenance, and discusses the challenges and problems faced by the current power robot standardization. Finally, combined with the international standardization trends and the development direction of power robotics technology, the future development of power robot standardization is outlook, put forward to strengthen international cooperation, improve the standard system, promote technological innovation and other recommendations, with a view to promoting the continuous progress of power robotics technology and the sustainable development of the power industry.

Based on topology optimization, a lightweight design for an intelligent transport robot's front claw, capable of automatic opening, closing, and slight deflection, was developed. The front claw was 3D printed using a 40% fill rate and polyester resin material, addressing the issues of the robot's slippage and line-tracking speed. ANSYS analysis was used to study the stiffness of the front claw, examining the deformation amount and positions in the Y and Z directions, and overall deformation for the front claws and support connecting rods made from ABS and polyester resin in both open and closed states. The deformation amounts in the Z direction and overall, with and without gravity, were compared, validating the stability and effectiveness of the front claw driven by rack and pinion engagement under various conditions, and confirming that the ABS connecting rod deformation meets the requirements for sensor functionality. The motor and servo linkage achieved motion control of the front claw. Through multiple tests, including opening claw backward rotation, slight deflection during opening and closing, stable line-tracking with an open claw, and black frame stability tests, it was repeatedly verified that the front claw of the intelligent transport robot meets the enterprise requirements for transporting a Rubik's cube from any position to the target area.

When diagnosing faults of different types of bearings in robots, there are not only differences in edge distribution between the source domain and the target domain, but also differences in internal classification. Considering the high requirements for stability and accuracy in robot bearing fault diagnosis, this paper proposes a neural network model named DAOCM based on multi-kernel MMD and domain adversarial for fault diagnosis of different types of bearings in robots. As a domain difference measurement method, multi-kernel MMD can simultaneously reduce the edge distribution distance and conditional distribution distance of the source and target domains. Domain confrontation can help transfer knowledge and patterns learned in the source domain to the target domain through domain adaptation, improving the generalization of the model. The experiment shows that DAOCM can achieve fault diagnosis of different types of bearings under variable working conditions. The accuracy of bearing fault diagnosis is above 96%.

With the rapid development of high-speed railway technology, the reliability of track circuit system becomes the key to ensure the safe operation of trains. ZPW-2000A track circuit system is the core component of high-speed railway signal system, its stability is directly related to the safety and efficiency of train operation. In this study, the reliability of ZPW-2000A track circuit system is evaluated by two methods: fault mode and impact analysis (FMEA) and fault tree analysis (FTA). Through identifying the main fault modes of the system, analyzing the impact of the fault and its causes, and using FTA method to identify the weaknesses of the system, this study put forward targeted recommendations for improvement.

Path planning is one of the key technologies for autonomous navigation of mobile robots. The A* algorithm is a commonly used algorithm for path planning. However, the traditional A* algorithm has problems such as too many path turns, too many traversal nodes, long path finding time and long paths. Therefore, this paper proposes an improved A* algorithm based on a bilevel extension strategy, including obstacle block-based extension and feature corner point-based extension. The algorithm divides the obstacles on the raster map into different blocks, the outer level algorithm is responsible for planning the connection direction between different blocks, and the inner level algorithm plans the connection paths between specific feature corner points. The heuristic function of the traditional A* algorithm is also improved by adding directional angle information. Finally, the simulation and comparison experiments of the path planning algorithm were carried out using C++ and Matlab software, and the results show that compared with the traditional A* algorithm in two different map environments, the number of turning points of the paths planned by the improved A* algorithm in this paper is reduced by 57.1% and 46.7%, the search time is reduced by 62.2% and 48.9%, and the total length of the path is reduced by 8.3% and 12.7%, respectively. The experimental results show that the improved A* algorithm can achieve shorter path length, fewer turns, and shorter search time, which greatly improves the performance of the traditional algorithm.

A simulation analysis was conducted on the steering of the multi-legged wheeled robot. The feasibility of the Ackermann steering principal usage in the multi-legged wheeled robot was analyzed in the paper. Through the co-simulation of ADAMS and MATLAB, the Ackermann steering principle used in the multi-legged wheeled robot was found out and improved. Based on the simulation results, experimental verification was conducted, and the results indicated that the Ackermann steering principle was suitable for multi-legged wheeled robots, but there were also many shortcomings to overcome in order to practical application.

In all position welding, the reasonable setting of welding process parameters at different positions and effective control of smooth transition are one of the difficulties in the automation of pipeline circumferential welding process. According to the process parameter requirements of all position welding, the welding trajectory is divided into spatial segments, and a cubic interpolation fitting algorithm is designed to achieve a smooth transition of welding parameters between each segment. According to the strategy requirements of parameter control, a PID control system was designed to improve the dynamic characteristics during the parameter adjustment process, achieving precise control of welding parameters and smooth transition.

Under complex working conditions, the reconnaissance robot with two degrees of freedom head has difficulty in target tracking due to its complex control system. According to the particularity of vehicle-mounted two-degree-of-freedom head, the tracked robot with tracking and reconnaissance is taken as the research object. Because the ground is uneven when the reconnaissance robot moves, the pitch Angle and azimuth Angle of the two-degree-of-freedom head will produce vibration, which will affect the tracking and aiming. Therefore, the Lagrange function method is used to establish the dynamic equation of the head, and then the dynamic model is established. Finally, the mechatronics co-simulation of the weapon platform is carried out by using matlab/Simulink module and Recurdyn dynamics software. The simulation results show that the control error of the PTZ is small and the precision is high, which verifies the feasibility of the control strategy and mechanical structure, and meets the design standard.

Path planning is a supporting technology in the field of industrial robot research. Appropriate path planning algorithms can ensure the safety of industrial robots at work and improve their work efficiency. This paper proposes an SRRT path planning algorithm based on Sarsa reinforcement learning, which retains the high efficiency of RRT node expansion and uses the feedback value of the expanded node as the evaluation criterion for the expansion process for subsequent learning processes. By improving the greedy strategy and reward function, the random tree follows the idea of "exploring first and then using", selects valid nodes with high feedback values as much as possible, reduces the amount of calculation in the subsequent learning process, improves the utilization rate of valid nodes, and expands the applicable environment of the algorithm.

Addressing the shortcomings in automatic deployment and retrieval as well as long-endurance capability of unmanned surface vessels (USVs) for inland waterways, this study focuses on the small-scale USV "Dolphin" as the research subject. A prototype of a docking station equipped with autonomous deployment/retrieval technology and wireless charging capabilities has been designed and implemented. Finite element analysis and dynamic simulation were employed to verify its feasibility. The results of practical application tests indicate that this prototype can further enhance the application capability and continuous operation time of USVs in complex environments, broadening their application prospects in fields such as marine scientific research, offshore patrols, and maritime rescue.

Acquiring terrain information during robot locomotion is pivotal for autonomous navigation, gait selection, and trajectory planning. Quadruped robots, owing to their biomimetic structures, demonstrate enhanced traversability over intricate terrains compared to other robotic platforms. Furthermore, the internal sensors of quadruped robots encompass rich terrain-related motion state data during locomotion across diverse terrains. This study investigates terrain perception in quadruped robots based on internal sensor data, probing the temporal signal relationships(including IMU data, body velocity data, foot-relative veloctiy data, and overall body velocity)influenced by motion state variations as quadruped robots traverse different terrains. By examining the interplay between terrain conditions, motion states, signal attributes, and locomotion patterns, we elucidate the mapping correlation between distinct terrains and internal sensor signal characteristics. This mapping correlation is acquired through the development of a deep learning model utilizing a self-attention mechanism. Employing a multi-label classification algorithm, we classify complex terrains and establish multiple physical feature labels—uneven, slippery, soft, and gradient—to depict terrain attributes. A feature-labeled terrain dataset is established by abstracting diverse terrain features across various terrains. Unlike semantic labels (e.g., grassland, sand, gravel) comprehensible only to humans, feature labels provide a more granular and precise terrain characterization, encompassing broader terrain attributes. Furthermore, a novel multi-sensor fusion model is proposed, utilizing waveform diagram as input, achieving an average classification accuracy of 98.49% across diverse terrains.

This project aims to develop an acupuncture robot capable of replacing manual treatments for specific points. It focuses on two main aspects: utilizing machine vision for spatial registration of the robotic arm, patient, acupuncture tools, and optical navigator to accurately locate acupuncture points, and implementing a needle insertion mechanism. Key to this is the trajectory algorithm that guides the robotic arm from its initial position to designated acupuncture points while avoiding obstacles.

In order to improve the safety of the multi-vehicle collaborative handling process and the flexibility of mobile robot scheduling, this paper optimizes the design of target formations for different types of mobile robots to carry heavy objects well. Rolling moment, combined with the handling strategy, a multi-vehicle intact heavy object handling model was established. At the same time, the energy-based static stability description method ESM (Energy Stability Margin) and the spacing between each short-circuit mobile robot are introduced to evaluate the team's search for the optimal form through an improved genetic algorithm, and the initial population generation method and selection methods, crossover and mutation methods; finally, the advantages and disadvantages of the form proposed in this article are proved through example verification and comparison with traditional genetic algorithms. The algorithm for solving multi-vehicle collaborative handling teams optimizes suction and perception problems.

The EEG-based rapid serial visual presentation (RSVP) brain-computer interface (BCI) paradigm has been widely used to achieve quickly target detection due to its potential in detecting an interest target from a rapidly refreshing letter or image stream. In these applications, single-trial RSVP detection accuracy is very important but it is often compromised by the low signal-to-noise ratio and and non-linearity of the EEG signal. In this study, we established a multi-brain collaborative BCI paradigm and designed a multi-brain cooperative RSVP detection (MB-RSVP) algorithm based on multi-variable linear regression models. Using this method, the single-trial classification accuracy of the three RSVP paradigms was estimated. Online experiment results indicated that the proposed multi-brain collaborative BCI-based RSVP framework effectively improves classification accuracy compared to single-subject RSVP while retaining real-time performance. Furthermore, offline experiment results showed that the proposed method improved the classification accuracy by 20.48% compared to single-subject RSVP experiments. The proposed multi-brain collaborative BCI-based RSVP framework provides a significant impact for driving EEG-based RSVP towards real-world applications.

Manufacturing companies nowadays are dealing with a huge amount of data in manufacturing process. However, they are facing the problem that these data collected from equipment and production management system often have unclear hierarchies and inconsistent representations. To improve the convenience of data processing, the paper proposed a method using knowledge graph to express hierarchical and internal relationships of manufacturing data. Also with trend and correlation analysis being conducted, potential patterns of these data were found and organized into knowledge. Finally, the summarized knowledge were added to the basic knowledge graph to accumulate knowledge in manufacturing process.

With the continuous advancement of technology and rapid development of industries, the demand for engineering talents will also continue to increase. Different from general universities and liberal arts courses, application-oriented universities serve as important bases for cultivating engineering talents and should actively explore and practice the cultivation of practical innovation ability driven by scientific research projects. Traditional teaching of basic theories cannot effectively stimulate students’ interest in learning. Therefore, this study explores the cultivation of practical innovation ability of engineering students in application-oriented universities based on scientific research project-driven approach. Taking the development and testing of virtual Try-On software algorithms as an example, the specific implementation process is detailed, which enhances students’ innovation ability, problem-solving skills in practical engineering issues, and fosters students’ spirit of teamwork collaboration.

Steady-state visual evoked potential (SSVEP)-based brain-computer interfaces (BCIs) have garnered significant interest from researchers because of their high information transfer rate (ITR). Nevertheless, the efficacy of conventional recognition techniques is constrained by the length of the experimental data set and the number of stimulus targets. Although some studies have made progress in improving the recognition performance of SSVEP by employing deep learning algorithms, there is still significant room for performance enhancement. To address the challenges of SSVEP in short-time multi-target recognition, Our proposal involves developing a neural network based on deep learning for SSVEP that combines the Perceiver architecture and Large Kernel Depth-wise separable convolution, called PLKDNet. The network consists of an input convolutional layer targeting multiple electrode channels, a series of large kernel depth-wise separable convolutional layers, a cross-attention and self-attention mechanism that incorporates the Perceiver layer and an output layer. We evaluate the performance of PLKDNet on different datasets and compare it to two mainstream deep learning models under different training sample conditions. The comprehensive experimental results demonstrate that the PLKDNet model performs excellently in handling the short-time multi-target sample recognition task, and its classification accuracy outperforms that of the comparison algorithms, showing the potential to advance the recognition performance of SSVEP-based BCI systems.

In the context of intelligence and experience, auditory interaction has become a research focus. To investigate the impact of auditory warnings on drivers' perceived urgency and their auditory preferences, this study describes tests conducted on subjects using various warning characteristics. In a laboratory setting, 45 test subjects completed evaluation tasks for auditory warning signals involving three timbres (xylophone, piano, flute), three sound types (pure tone burst, continuous pure tone, melodic sound), three frequencies (500 Hz, 1250 Hz, 2000 Hz), and three rhythms (120 BPM, 60 BPM, 40 BPM). The evaluation included assessments of perceived urgency and subjective preference. Non-parametric tests were conducted on the experimental data, and correlation analysis was used to investigate the relationship between perceived urgency and auditory preference. Findings reveal significant influences of timbre, sound type, frequency, and rhythm on perceived urgency. Timbre, sound type, and rhythm also significantly impact drivers' preference scores. Faster rhythms increase perceived urgency, while slower rhythms decrease both urgency and preference. Flute timbre and continuous pure tone are perceived as most urgent but are less preferred by drivers.

The 3D morphology of rough surfaces is crucial for the study of the mechanical properties of bolted joint surfaces. This article uses fractal statistical methods to discuss the distribution patterns of characteristic parameters that characterize micro convex bodies. This model combines the fractal theory's ability to describe rough surfaces with the randomness introduced by Monte Carlo methods. Based on the study of rough surface morphology, the following conclusions can be drawn: fractal dimension D and feature scale coefficient G are negatively correlated with the contour of rough surfaces. As the fractal dimension D increases and the characteristic scale coefficient G decreases, the fluctuation degree of the micro convex body decreases, and the surface becomes smoother. In addition, the fractal dimension D has a more significant impact on surface contour compared to the feature scale coefficient G.

This paper mainly focuses on the information security threats faced by automotive CAN bus, and constructs a framework based on threat analysis and risk assessment, aiming at identifying, assessing and mitigating the impact of these threats on automotive network security. The framework based on threat analysis and risk assessment CAN effectively improve the information security level of automotive CAN bus, which has important theoretical and practical significance for automotive network security management. Through the experiment of attack recognition rate, the research technique achieves 90%, 85% and 95% high recognition rate under replay attack, fuzzy attack and DoS(Denial of service) attack respectively. It CAN be seen from the data conclusion that the framework based on threat analysis and risk assessment has high effectiveness and practicability in CAN bus information security protection.

In the field of Brain-Computer Interfaces (BCI), while research on Motor Imagery (MI) decoding models has advanced significantly, systematic studies on dataset construction and partitioning strategies remain scarce. Inappropriate data partitioning may lead to models performing well in experimental settings but underperforming in real-world applications. This study compares the impact of various EEG dataset construction and partitioning strategies on MI decoding model performance to optimize BCI systems. Utilizing the BCI Competition IV 2a dataset, we designed three data construction strategies: Intra-session training and testing, Cross-session evaluation, and Combined session evaluation. We applied four data partitioning strategies: holdout, cross-validation, shuffled cross-validation, and time series cross-validation. For model architecture, we employed the end-to-end deep learning framework EEGNet. Results indicate that increasing data volume enhances model performance. Among partitioning strategies, shuffled cross-validation showed superior performance in improving model accuracy. The holdout method also performed well with sufficient data, offering a computationally efficient alternative. These findings provide valuable insights for optimizing BCI system development and evaluation methodologies.

In congested waters, when an unmanned surface vessel (USV) is faced with a possible collision with multiple vessels, its navigation has to select a reasonable collision avoidance path within its own kinematic constraints. Therefore, to solve with this complex problem, a collision avoidance method based on the improved dynamic window approach (DWA) is proposed. First, a simple A* algorithm is applied to generate the global path and extract the key nodes. Next, the encounter model of the USV and the encountered ship was established, and a new evaluation rule is introduced on the basis of the COLREGs rule. In response to various encounter situations, the USV will take different actions to select the optimal local navigation path, thereby achieving successful collision avoidance. Upon successfully navigating through local obstacles, the USV seamlessly returns to its original global path. Finally, simulation experiments confirmed the algorithm's effectiveness and feasibility, demonstrating its practical significance.

In the research area of the Dynamic Flexible Job-shop Scheduling Problem (DFJSP), real-time responsiveness to disturbances and solution quality are the key factors of dynamic scheduling performance. Traditional methods can not offer ideal dynamic scheduling performance as they either lack real-time responsiveness to disturbances or sacrifice solution quality. In contrast, scheduling methods based on Reinforcement Learning (RL) allow the agent to interact with a dynamic environment and update the scheduling strategy in real-time, resulting in high real-time responsiveness and strong adaptability. However, relying solely on dynamic decision-making experience will limit the quality of the solution. Therefore, this paper proposed a novel Dual-Phase Adaptive Learning Framework (DPALF) that introduces a pretraining phase with Imitation Learning (IL) to equip the decision-making agent with the static scheduling capability before RL in the disturbed scheduling environment. Experimental results show that introducing the pretraining phase improves learning performance and enhances the stability of the formal training process. Additionally, the proposed method outperforms three representative RL methods in solution quality, proving more effective for real-time dynamic scenarios.

The current assembly process planning of complex precision products is complicated and time-consuming, and there has been an underutilization of assembly knowledge from previous product cases. To solve this problem, a method for reusing the assembly process of microdevices based on deep learning is proposed. By constructing the design structure matrix (DSM) of the product, the assembly relationship is expressed using the way of graph structure. Node features that refer to multi-dimensional attribute information of the parts and the graph are simultaneously fed into a graph neural network, which can determine the similarity score of any two assembly connection diagrams with extensive training of the model. Then the new products can reuse the assembly process of products with higher scores. On the other hand, in the absence of a similar product case for reference, the similarity between the local structures of the new product and previous products can be identified through a mining algorithm based on Maximum Clique. The assembly information from previous product cases can be fully utilized to guide the assembly process planning for the new product, thereby enhancing the efficiency of the process planning.

The scheduling problem of workshop production process is an important research object of intelligent manufacturing. In this paper, the idea of introducing reverse learning is integrated into NSGA-II algorithm, and an improved NSGA-Ⅱ algorithm based on reverse learning is proposed to solve the P-FJSP problem, which is easy to prematurely converge to the local optimal solution. It is especially suitable for the production scheduling problem of large quantities and batches, and it is also proved that the algorithm can be well applied to the production line scheduling system.

Indirect Reciprocity (IR) is the main mechanism to promote cooperation. A multi-agent system is proposed in this paper to simulate IR with optional interactions so that more specific experiments could be carried out. Comparative experimental results verify the rationality and feasibility of the simulation model structure, and more parameters’ effects on the optional IR are further analyzed.

The deployment of base stations (BSs) in 6G will be much denser and more flexible. Thus, integrated access and backhaul (IAB) networking has been widely studied in 6G. Existing IAB networking is based on the directed-acyclic graph (DAG) architecture. In this architecture, two neighboring IAB nodes have a fixed parent-to-child relation, which highly limits the flexibility of link scheduling. In this paper, a mesh-based scheduling scheme (MeshABS) is designed to eliminate the limits of the DAG scheme. The separated DU-MT functions in the control plane are replaced by one FU function and a configurable relation for two neighboring IAB nodes. A joint scheduling algorithm for mesh-based IAB (JointMIAB) is developed to conduct link scheduling, mode configuration, and routing for each IAB node jointly. The max-min fairness on throughput among all IAB nodes is achieved. Simulation results show that MeshABS can improve the throughput by 4.1%-73.3%, and compared to existing schemes, MeshABS substantially reduces end-to-end delay.

In the process of processing large-scale maps with complex environment information, JPS jump point search algorithm has many ineffective expansion nodes, blind search direction and search scope, which leads to long pathfinding time and large memory cost, and so on. The two improved algorithms introduce the definition of standard center vector and constrained region, specify the priority of the jump point to be expanded in the direction according to the degree of deviation between the direction to be expanded and the target direction, and delimit the constrained region to distinguish the expansion priority of the jump point to be expanded in the region, and comprehensively judge the priority order according to the direction priority and regional priority. The simulation results show that the path finding time of the two improved JPS algorithms is significantly shorter than that of the traditional algorithms, and the number of extended invalid jump points is significantly reduced. Therefore, the improved algorithm can significantly improve the efficiency of pathfinding, and it has more application significance than the traditional JPS algorithm.

This article delves into the technical essentials of porting and developing CSI interface camera drivers in embedded Linux systems. It begins by comprehensively and deeply exploring the principles of camera controller driver under the V4L2 video capture framework. Subsequently, the article elaborately describes the design strategies of camera drivers. Following that, successful porting of the OV2640 camera with CSI interface was achieved on the NXP i.MX6ULL development board, and image capture and video acquisition tasks were smoothly executed. Through careful examination of image attributes, it was confirmed that the captured images fully met the predetermined performance standards.

Chu music and dance originate from the ancient Chu state in China, serving as a crucial component of Chu culture. They exhibit the unique charm of Chu music culture and carry forward the historical memory and cultural traditions of the Chu state for future generations. However, in today's society, traditional Chu music performances face limitations in dissemination. Additionally, the traditional allure and interactivity of Chu music performances are insufficient, making it challenging for audiences to deeply immerse themselves. There is a pressing need to establish a more intimate interactive connection with Chu music and dance.With the widespread application and development of image recognition technology, its utilization in the field of dance design has become increasingly prevalent. Therefore, this paper proposes a design for an interactive installation for Chu music and dance based on image recognition. The aim is to create an immersive environment for Chu music and dance, providing users with a tactile interactive experience. This design endeavor seeks to bridge the gap between existing Chu music and dance performance forms and the demands of modern audiences. By enhancing immersion and interactivity, users can more actively participate in Chu music and dance performances.

As a new form of human-computer interaction, brain-computer interface (BCI) has been widely studied. However, most traditional BCI paradigms still face issues such as high cognitive load and easy fatigue. In recent years, intuitive BCI paradigms such as speech imagery (SI) and visual imagery (VI) have gradually become new research hotspots. Unfortunately, the classification performance of standalone SI and VI is still not ideal. In this paper, we designed an intuitive BCI paradigm combining SI and VI (SI+VI). Subjects are asked to simultaneously imagine images and Chinese characters associated with specific drone flight actions. Additionally, a dual-branch network using cross-attention mechanism to fuse time and frequency features (TFCA-DBNet) of electroencephalogram (EEG) signals is proposed. Experimental results show that the proposed TFCA-DBNet achieves a highest accuracy of 82.33% and an average accuracy of 67.33% in the four-class SI+VI task, which is improved by 15.80% and 6.90% compared to SI and VI, respectively. This indicates that the SI+VI paradigm has higher distinguishability than that of SI or VI alone. Compared with other methods, TFCA-DBNet also achieves better classification performance, which proves the effectiveness of the proposed method. Our study provides valuable reference and guidance for future applications of intuitive BCI paradigms.

Two-dimensional continuous control based on brain-computer interface (BCI) systems is becoming a research hotspot. However, due to the discreteness in BCI signal decoding, achieving smooth 2D continuous control is often challenging. To address this issue, this paper proposes an innovative 2D continuous control method that converts three discrete commands into continuous ones and validates it through a real-time online BCI system. We used dry electrodes for signal acquisition and applied the Canonical Correlation Analysis (CCA) algorithm for electroencephalography (EEG) decoding. Two simulation experiments were designed: obstacle avoidance for unmanned vehicles and trajectory tracking for unmanned vehicles, to verify the effectiveness of the 2D continuous control method. By analyzing experimental data from three subjects and multiple evaluation metrics, we comprehensively and thoroughly assessed the proposed method.

The aim of human-robot collaboration (HRC) is to efficiently utilize robots' consistent operations and human cognitive abilities in handling complex assembly tasks. Accurate comprehension of human intentions is pivotal for successful collaboration. Existing methods typically focus on recognizing a closed set of actions, making it challenging to predict all action categories in an open environment. Such methods often misrecognize human intentions by wrongly recognizing actions outside the training set as known categories, resulting in safety concerns. To address this challenge, we proposed a novel approach for action recognition in open environments in human-robot collaborative assembly. This method ensures the accuracy of known action recognition and notably reduces misrecognition rates of unknown action categories. It employs a spatial-temporal graph convolutional networks (ST-GCN) model based on skeleton features for real-time and robust action recognition. Additionally, the OpenMax algorithm is introduced to enhance the ST-GCN model, enabling the recognition of unknown action categories. Extensive experiments on custom and public datasets demonstrate that the improved algorithm effectively recognizes unknown action categories, outperforming traditional methods.

Assembly of a complex product varies from variety to batch production mode, and the assembly process preparation of multi-model batches is extremely tedious, and the sharing and reuse of assembly process knowledge is poor, and the similar assembly process contents of different models are still repeatedly prepared. The structure of different models of the same series of products is similar, and the assembly process knowledge can be referred to or reused in the process design of the same series of products. In this paper, an assembly process knowledge reuse method is proposed, and an assembly process knowledge graph embedding model based on hierarchical attention network HAGAT is proposed. Nodes and edges in the assembly process knowledge graph are mapped to low-dimensional vector space, and the reusability of the existing assembly process is judged according to the similarity of the graph pairs between the new product and the existing product. Taking the assembly process design of a complex product as an example, the assembly process content of a new product is supplemented on the basis of the existing assembly process knowledge graph database, so as to realize the automatic reuse of assembly process knowledge. The experimental results show that the effectiveness and performance of the proposed method are superior to the existing methods, and it can effectively improve the efficiency of the product assembly process design.

In this study, we explore the integration of emotional expression in oil paintings with virtual reality (VR) technology, focusing on enhancing audience art experiences through enriched virtual interactions. Vincent van Gogh's "La Mousmé" serves as our subject of investigation, where VR technology is used to simulate emotional and environmental interactions depicted in the oil painting, aiming to offer a novel mode of art appreciation. The objective of this research is to expand the expressive potential of traditional oil paintings through technological means, enabling audiences to directly interact with artworks in virtual environments to gain deeper insights into the artist's emotions and creative intent. By integrating visual, auditory, and even tactile elements into the virtual experience, this project not only enhances the emotive power of oil paintings but also provides an innovative tool for art appreciation, facilitating a better understanding and appreciation of oil painting culture among audiences. Furthermore, by constructing a virtual space that simulates the artist's creative process, audiences can freely explore and experience every step from initial sketches to the completed work, gaining profound insights into the complexity and allure of artistic creation. This paper aims to provide new perspectives and practical approaches for the digital inheritance and innovation of oil painting art through the application of these technologies.

Due to remote orbital maneuvers, payload separation and fuel consumption, the mass properties and dynamic parameters of gravity satellites change in real-time. These changes make it challenging to maintain high-precision stability in attitude control of the spacecraft, affecting the execution of space missions. This paper establishes an attitude dynamics model for gravity satellites that explicitly includes on-orbit variable parameters. Furthermore, backpropagation (BP) neural network is used to identify the dynamic parameters of gravity satellites. Finally, BP neural network is compared with the recursive least squares algorithm (RLS) and the random weight particle swarm optimization (PSO) algorithm. The results indicate that the identification error mean of the BP neural network can achieve 10^-3 orders of magnitude, and the root mean square error can achieve 10^-1 orders of magnitude. The identification accuracy and robustness are superior to those of the RLS algorithm and the random weight PSO algorithm. This paper compares the different high-precision on-orbit identification algorithms for dynamic parameters of gravity satellite, laying a foundation for the identification algorithms design and selection of the next generation gravity satellites.