Table of contents

The 2023 4^th International Conference on Internet of Things, Artificial Intelligence and Mechanical Automation (IoTAIMA 2023) was held via hybrid form in Guangzhou, China during July 21^st to 23^rd, 2023, following the successes of previous events held in Hangzhou (2020, 2021), and Shanghai (2022). This year the IoTAIMA 2023 aimed to bring together researchers, developers, and users in both industry and academia in the world to discuss emerging issues on Internet of Things, artificial intelligence and mechanical automation, for sharing state-of-art results, and for exploring new areas of research and development.

The Conference was organized by Peking University-Wuhan Institute for Artificial Intelligence and co-organized by Central South University School of Automation. More than 80 delegates from home and abroad attended the Conference.

The IoTAIMA 2023 was featured with four keynote speeches (by Prof. Deyu Qi, Prof. Jinping Ao, Prof. Ming Jiang, Assoc. Prof. Wei Wei, respectively), and several oral and poster presentations, in which a wide range of topics were covered and the most recent significant results were presented. Prof. Ming Jiang from Sun Yat-sen University, China addressed his speech Massive MIMO Aided Multinetwork 3D Positioning. Exploiting a distributed M-MIMO framework, his team proposed to employ a deep belief network (DBN) to analyze the received signal strengths (RSS) generated by a diffraction model. Next, the preliminary DBN estimates were forwarded to a long-short term memory network (LSTMN), where the trajectory information of the targeted UE can be extracted based on much less historical trajectory information than existing solutions. Then, the three-dimension (3D) coordinates of the UE's positions can be estimated with a back propagation neural network (BPNN) which combines the outputs of DBN and LSTMN. Finally, extensive simulation results were provided to demonstrate the effectiveness of the proposed BPNN scheme.

Quantities of excellent papers evaluated based on their originality, technical or research content, correctness, relevance to conference, contributions, and readability were presented in the Conference Proceedings published by Journal of Physics: Conference Series (JPCS). The topics of these papers include Radio Frequency Identification, Micro-Electro-Mechanical Systems, New Sensing Technology, Unmanned System Control Technology, Industrial Robots and Automatic Production line, etc.

With the excellent quality of all the presentations, the IoTAIMA 2023 was a great success. We wish to thank the sponsors of the Conference, and particularly the Technical Program Committee. We would also like to extend our gratitude to all the speakers and authors for sharing scientific ideas and presenting new perspectives with us on related topics. Our appreciation also goes to the editors and other staff of JPCS for the assistance it provided for the publication of this paper volume.

The Committee of IoTAIMA 2023

List of Committee Member is available in this Pdf.

All papers published in this volume have been reviewed through processes administered by the Editors. Reviews were conducted by expert referees to the professional and scientific standards expected of a proceedings journal published by IOP Publishing.

• Type of peer review: Double Anonymous

• Conference submission management system: Morressier

• Number of submissions received: 88

• Number of submissions sent for review: 85

• Number of submissions accepted: 51

• Acceptance Rate (Submissions Accepted / Submissions Received × 100): 58

• Average number of reviews per paper: 2

• Total number of reviewers involved: 22

• Contact person for queries:

Name: Fengxin Cen

Email: alan.cen@gsrassn.org

Affiliation: Global Scientific Research Association

It is a significant concern that there is a risk of passenger intrusions at station platform ends. Current detection uses video triggered by single-line radar, but it is ineffective for accurate identification. In this paper, we address this issue by first analyzing the characteristics of intruders at the ends of train platforms. We propose a two-stage filtering-recognition method to achieve intruder detection based on single-line radar point cloud data. In the first stage, we smooth initial point cloud data using a double-chain exponential weighted moving average filter by grouping points. In the second stage, we extract features using the background subtraction method and a critical threshold of point numbers to detect intruder targets. Experimental results demonstrate that this method is effectively capable of detecting intruders at different distances.

Artificial intelligence and robotics are increasingly used in chemical experiments for reagent preparation, but it is still a challenge to accurately grasp the reagent bottle in small and dense scenarios. Therefore, precise positioning and identification of the reagent bottles are crucial. However, the different kinds of chemical reagents create complexity in target detection, making the task difficult. Therefore, we propose a model, named TA-YOLOv5, based on YOLOv5 for the task of chemical reagent localization. To enrich our dataset and prevent single-background overfitting, we employed Mosaic and Mix-Up data augmentation during dataset processing. Furthermore, Transformer Encoder and Coordinate Attention modules are utilized to enhance the feature expressive ability for both small target objects and global information during model training. Finally, the CIoU-NMS algorithm is utilized to optimize prediction box filtering in dense scenarios. In addition, for the task of chemical reagent classification, this paper uses the CRNN (Convolutional Recurrent Neural Network) algorithm to identify character information on the label of the reagent bottle, enabling the successful classification of the reagent bottle. The experimental results illustrate that TA-YOLOv5 achieves 98.49% precision on the self-made reagent bottles dataset, which is 4.12% higher than the YOLOv5 network model. Moreover, the CRNN network reached 97.6% accuracy on IIIT 5k dataset.

The central pattern generator (CPG) is a micro circuit in neural system and it can generate rhythmic signals to regulate locomotion. The researchers have investigated the features of the CPG, and they have paid more attentions to the the programmable characteristic. In this address, a new learnable CPG based on Wilson-Cowan oscillator is established. The sine signal, the complex signal, the chaotic signal and angles of compass-like robot are used as input to test the new programmable central pattern generator. The simulations present that the learnable CPG has the ability to learn different signals effectively. These results are the significant contribution to the research of the programmable CPG.

This paper designs a pipelined memristive neural network ADCs. Cascade the sub stages of memristive neural network ADC with pipeline structure to improve the conversion accuracy of memristive neural network ADC. First, the signal flow of the pipeline architecture is optimized, and the compatibility between the 4 bit memristive neural network ADC and the pipeline architecture is solved. Secondly, the application of random disturbance circuits in pipeline architecture was analyzed. Combining the calibration circuit with the pipeline DAC to reduce the power consumption of the calibration circuit. Finally, the circuit model of memristive neural network ADC is built and compared with the existing memristive neural network ADC. The results indicate that the pipeline structure ADC designed in this chapter has the advantage of adaptive calibration in terms of calibration function.

Registration is a critical task in the field of point clouds, aiming to align data acquired at different times or from different viewpoints for accurate matching. Deep learning methods have made important progress in point cloud registration tasks. However, most existing approaches do not handle the non-overlapping parts of point clouds, resulting in poor performance in low-overlap and noisy scenarios. We propose a registration model called OPSNet, which achieves optimal alignment transformation estimation and overlapping region prediction through an iterative process. OPSNet consists of modules including global feature extraction, overlapping region prediction segmentation, and alignment registration. By utilizing a segmentation algorithm to deal with the non-overlapping parts of data, OPSNet reduces the adverse effects caused by non-overlapping regions in point cloud registration. The model learns feature representations and performs iterative optimization to achieve precise point cloud alignment. We conduct comprehensive experiments on common point cloud registration datasets and compare OPSNet with several classical point cloud registration methods. The experimental results display that OPSNet achieves outstanding performance in terms of rotation and translation errors, outperforming other methods. Additionally, we evaluate the registration performance under different overlap ratios and find that OPSNet can achieve better registration results even in low-overlap scenarios.

To make use of past practical experience and expertise effectively and accurately, and reduce the waste of resources due to repetitive labor, this paper conducts intricate multi-hop logical deduction by using the information obtained from the process knowledge graph (PKG). Here, BoxE is introduced in this context, which is a box embedding framework designed to address diverse First-Order Logic queries (FOL) across Process Knowledge Graphs (PKG). A key insight of BoxE is that entities and queries can be embedded as boxes and geometric operations can be used iteratively for logical reasoning to answer the complex logical queries. Our demonstration showcases BoxE's ability to effectively respond to various First-Order Logic queries involving technology, process, procedure, etc. on PKG and shows that BoxE has higher accuracy on each query structure, which can generalize unseen query structures and answer complex query structures scalably and accurately.

To solve the problem of low data processing rate in traditional medical information storage methods, a safe centralized storage method of medical big data information is proposed. The data set storage model is designed through the Internet of Things, and VeritasBackupExee is used as data recovery and backup software, and data recovery and backup are realized by the Internet of Things. Encrypted storage of ciphertext index vector is carried out by using the ciphertext storage algorithm, and the medical data center of Internet of Things middleware is constructed. By sharing storage devices with multiple hosts, highly reliable backup can be realized. The average response time of the method in this paper is different for medical big data services. The average response time for modifying service is 11ms, creating service is 12ms, and migrating service is 13.7 ms. To verify the data service response time of this method, it is necessary to verify the response performance of different methods. The results show that this method has higher data download and upload rates and faster data service response.

Aiming at the problems of frequent packet loss and high energy consumption in large-scale LoRa networks, this paper proposes a joint allocation method of Spreading Factor (SF) and Coding Rate (CR). Firstly, we establish the effect relationship of SF and CR on Frame Error Rate(FER), and pre-allocate SF and CR based on the node position to minimize the energy consumption of the node while meeting the FER requirement. Then, according to the collision model, the collision probability of each SF group is calculated, and the sequential water injection method is used to equalize the collision probability within each SF group, to improve the average packet arrival rate of the entire network. The simulation results show that compared to mainstream algorithms, the proposed algorithm obtains a 13% increment in terms of the network average packet success probability, and a 104% one in terms of the average energy efficiency. The proposed algorithm has high application value in many application scenarios such as smart agriculture and smart cities.

In this paper, the bi-objective AUV path planning problem considering the shortest path and the minimum path smoothness is studied. Firstly, a mathematical model of the problem is established and a two-dimensional Cartesian coordinate system model is used to model the AUV working environment. Then the dual objective is normalized and transformed into an adaptation function by dimensionality reduction. An improved Genetic Algorithm is proposed to solve the problem according to its characteristics. An initial solution generation strategy for the n-1 equipartition method based on the improved oscillation factor is designed, and an elite retention strategy with simulated annealing operation is introduced to improve the genetic strategy. To verify the effectiveness of the improved algorithm in this paper, simulation experiments were conducted by the Microsoft Visual Studio platform. After the comparison and analysis of the three algorithms, the algorithm proposed in this paper has shorter and smoother paths, faster convergence speed, and more obvious advantages in the quality of the solution.

The problem addressed in this paper is the increase in the parameter quantity of existing human body pose estimation network models when the depth of the network is scaled up to improve forecast precision. To tackle this issue, an optimization network model of the new human body pose estimation called BDENet is proposed, inspired by high-resolution detection networks. This model incorporates a bottleneck structure and dilated convolution to reduce parameters and incorporates the ECA lightweight attention mechanism to enhance precision. Compared with HRNet, the proposed model achieves a 21.4% reduction in parameter quantity on the MSCOCO dataset while scaling up precision by 1.4%. The experimental findings strongly suggest that the modified network significantly improves network precision along with lowering the number of parameters.

In action recognition based on the skeleton, graph convolutional networks (GCNs) have shown great superiority based on the design of making the skeleton in the video a spatiotemporal map and extracting features from the spatiotemporal map. However, the topology of the skeleton in GCN-based methods is pre-designed according to prior knowledge, which limits the capacity of the network to learn high-level topology about the skeleton. To improve this deficiency, we design a temporal-difference adaptive graph convolutional network (TDA-GCN) that can learn the potential topology of the human skeleton from the input data, which is augmented using the channel attention module. Experiments show that TDA-GCN achieves state-of-the-art performance on two large-scale skeleton datasets, NTU-RGBD and Kinetics-Skeleton.

This paper proposes a research method to enhance the accuracy and real-time capability of helmet detection in complex industrial environments, aiming to address the engineering challenges of poor robustness and significant occurrences of both false positives and false negatives in existing detection methods. In this study, the C2F (faster version of CSP Bottleneck with two convolutions) module and FE (FasterNet with EMA) module are integrated into the network architecture of YOLOV8 to form a new attention mechanism module called C2F-FE. This module enhances the model's perception of safety helmet targets by fusing feature information from different levels and incorporating attention mechanisms while reducing computational overhead. Furthermore, the model is trained and optimized on publicly available safety helmet datasets. Experimental results demonstrate that the improved model exhibits stronger robustness, achieving an accuracy rate of 94.6% and a mAP50 of 99.1% for safety helmet detection in complex construction scenarios, with an inference time of 0.7 ms.

Insulators are important components for achieving electrical insulation and mechanical support, but they are prone to various defects in harsh operating environments, which can damage their mechanical strength and insulation performance. This article proposes the Shuffle YOLOv7 model based on the YOLOv7 algorithm for insulator defect detection, aiming to solve the weakness of low precision in traditional object detection algorithms when facing complex backgrounds and small-sized defects. To address the issue of low attention to flashover faults in traditional algorithms, the ShuffleAttention fusion attention mechanism is supplied to concentrate on both intra-channel and inter-channel deep features, and the original PANet structure is replaced with a pyramid which has a bidirectional feature fusion structure to enhance the network's feature extraction ability. The Focal-EIOU LOSS optimization method focuses on high-quality prior boxes to improve model accuracy, and the effectiveness of the optimization method is verified through ablation experiments. These results of the experiment show that the proposed algorithm achieves varying degrees of performance improvement in terms of precision, recall, average precision, and overall loss compared to mainstream object detection algorithms in detecting insulator damage and flashover.

In view of the long training time for the optimization of the network model parameters of the SELM and the uncertainty of the model generalization ability, this paper proposes an analog circuit fault diagnosis model based on the sailfish algorithm to optimize the stacked kernel extreme learning machine(SKELM). This model introduces a kernel function to build a multi-layer KELM, which can improve the generalization ability and learning speed of the feedforward neural network. The weights of each layer of SKELM are obtained through the automatic encoder training based on the KELM. Since KELM-AE does not need to set initial parameters, the training speed is improved. However, the kernel parameters and regularization coefficients of KELM-AE are set manually, so the sailfish optimizer (SFO) is used to optimize these two parameters, and then the optimal SKELM model is built through layer by layer training. Finally, the Leap frog filter circuit is used as the simulation experiment circuit, and further compared with the optimized SELM. The results show that KELM-AE has strong generalization ability, and it can map fault features to high-dimensional feature space through nonlinear mapping without extracting fault features separately, thus improving the classification accuracy.

With the rapid development of suburban rail transit, fire safety issues have received more and more attention. As the first line of the Shanghai Suburban Railway, the fire safety problem of the Shanghai Suburban Railway Airport Link Line is more prominent. In this paper, an automatic fire alarm system is designed and researched for the fire safety problem of the Shanghai Railway Airport Liaison Line, which adopts a hierarchical distributed structure and can realize accurate fire detection thorough immediately sending fire alarm information to the station control room and line operation control center and informing the location of the fire area. At the same time, the system also coordinates with the BAS system and ISCS system or independently realizes the linkage control of fire fighting equipment. The research results of this paper are of great significance for improving the fire safety level of Shanghai railway airport links.

In this paper, the system of dynamic identification and monitoring of water bodies and intelligent allocation of pharmaceutical discharge is designed. At present, aquaculture technology in our country is relatively traditional, and the water environment-bearing capacity will often be ignored. This paper aims to realize the efficiency of equipment in the multiple stages of aquaculture through the design system, the remote control of motion software device, and strive to combine intelligent equipment and the basic process of aquaculture, so as to make the aquaculture industry gradually upgrade. Through the test, our device can dynamically identify and detect the water body, and ensure the fishery output and water quality at the same time, bringing economic and environmental benefits to a great extent.

The intelligent vehicle designed in this paper can realize functions, such as safety detection, visual identification, remote control and manipulator grasping, and so on. Arduino MEGA is used as the main control board to send signal messages to drive vehicles. Wi-Fi module is used to receive messages to remote control vehicles. The ultrasonic and infrared module is used to realize object detection around vehicles. To realize complex route movement, raspberry pie is used for visual recognition and path planning. Data is sent to Arduino for judgment in real time. Finally, it is verified that the design effectively improves the path-planning ability and obstacle-avoidance function in a sample vehicle.

Due to the differences in the line resistance of each distributed generation unit, conventional sag control does not allow for precise distribution of power, which also reduces the problem of system reactive power accuracy. To address this problem, this paper analyzes the power equalization and distribution conditions. A method for adaptive adjustment of virtual resistance by system output power feedback is submitted. Reducing the variance among the original line resistances aims to solve the problem of power not being evenly divided and improve the equalization of reactive power. By building a Simulink parallel simulation model of the inverter, it is verified that the power feedback-based fetching and the improved droop control is effective.

In recent years, with the development of sensors, communication networks, and deep learning, drones have been widely used in the field of object detection, tracking, and positioning. However, there are inefficient task execution and some complex algorithms still need to rely on large servers, which is intolerable in rescue and traffic scheduling tasks. Designing fast algorithms that can run on the airborne computer can effectively solve the problem. In this paper, an object detection and location system for drones is proposed. We combine the improved object detection algorithm ST-YOLO based on YOLOX and Swin Transformer with the visual positioning algorithm and deploy it on the airborne end by using TensorRT to realize the detection and location of objects during the flight of the drone. Field experiments show that the established system and algorithm are effective.

Aiming at the problems of cumulative error in monocular visual positioning and Non-Line-of-Sight (NLOS) error in UWB positioning for the automated guided vehicle (AGV) in indoor environments, a combined method of vision and Ultra-Wide Band (UWB) is proposed for indoor AGV positioning. Firstly, the overall structure and system of the AGV are designed to achieve indoor navigation and positioning functions. Secondly, the monocular visual and UWB positioning data are fused using the Error State-Extended Kalman Filter algorithm (ES-EKF) to obtain the optimal pose estimation of the AGV. Finally, the AGV is used as a mobile platform to conduct positioning experiments in different indoor environments. The experimental results demonstrate that the navigation and positioning system has high accuracy and robustness in indoor environments with obstacles, and no significant drift or discontinuity phenomena occur during the positioning process, indicating its practicality in indoor settings.

Given the actual situation of greenhouse production nowadays, this paper proposes a kind of intelligent light replenishment system for greenhouses based on a Bluetooth Mesh network. The system uses the ESP32-C3FH4 microcontroller from Espressif as Bluetooth Mesh nodes to form a Bluetooth Mesh monitoring network. This paper designs an intelligent IoT device control system based on Bluetooth Mesh network technology and cloud platform services. Through the flooded network composed of Bluetooth Mesh, the natural light intensity data monitored by each service node is sent to the cloud platform periodically through the gateway, the monitored data is accessed by the cloud platform for management, and the artificial light source module is triggered according to the deposited data to realize automatic control, realizing the modernized digital management function of agricultural greenhouses.

The precise construction of the master-slave control strategy is a fundamental requirement for robot-assisted surgery. In this study, we propose a master-slave control strategy for gastrointestinal surgical robots based on the natural orifice transluminal endoscopic surgery surgical robot developed by Tianjin University. The forward and inverse kinematics of the continuum mechanism at the slave manipulator are investigated. A master-slave coordinate mapping system for surgical robots is established by analyzing the kinematic characteristics of the master and slave manipulators. Additionally, an intuitive mapping algorithm and a control flowchart for the robot are developed. The proportional control and incremental control strategies tailored for natural orifice transluminal endoscopic surgery procedures are proposed based on intuitive mapping control. Finally, a master-slave trajectory tracking experiment is performed to validate the feasibility of the proposed master-slave control strategy.

The phenomenon of elderly people living alone was becoming increasingly common in the new era, so the demand for monitoring the living environment of elderly people living alone was becoming increasingly strong. The design of home environment monitoring for elderly people living alone was designed based on a cloud platform. The system mainly consists of an STM32 main control module, a temperature and humidity detection module, an Internet of Things module, a security module, a relay control module, and a user App terminal. Then the fuzzy PID algorithm was designed to remotely control indoor temperature and humidity. After practical operation verification, the system can perform local and remote monitoring of the living environment of elderly people living alone and has the characteristics of reliability and stability.

This paper presents a novel approach for evaluating the pronunciation quality of English speech using continuous speech recognition technology. The research focuses on the application of artificial intelligence in speech recognition, utilizing web browsers on various terminal devices such as computers, mobile phones, and tablets to allow users to read the provided text aloud. The web program captures audio input from the microphone, records it in MP3 format, and uploads it to the server. The server employs the Whisper model to transcribe the audio into semantic text, which is then compared with the displayed text. By calculating the semantic distance and assessing the accuracy of pronunciation, the system provides an evaluation of pronunciation quality, marking correct and incorrect words. To achieve real-time processing, the compact tiny model is employed, and further optimization is performed using Ctranslate 2, resulting in significant performance improvements.

Multi-pedestrian tracking is one of the hot topics in computer vision. For an intelligent mobile robot, multi-pedestrian tracking from a first-person perspective can provide information for navigating through a crowd and ensure safety. Most of the existing methods cannot deal with occlusion and trajectory overlap well. In this paper, a multi-pedestrian tracking method fusing two-stage matching is proposed. Firstly, the detection and the corresponding feature values of the pedestrians are obtained by a multi-task learning network based on CenterNet. Then the detected pedestrians are matched with feature values by greedy strategy. When dealing with the reappearance of pedestrians caused by occlusion or trajectory overlap, the sample database is established to update the samples in real time. The color histogram and HOG feature are calculated for each sample. When the pedestrian disappears, the direction of disappearance and the last position is recorded for the selection of trajectory. Finally, the KM algorithm is used for cross-frame matching. Our method is compared with some recent methods on MOT data sets. The result shows that our method has a significant improvement in the main evaluation index MOTA.

The paper proposes the Face and Background Deepfake Detection (FBDD) algorithm to detect both face forgery and background forgery. Multiple key frames of the video are extracted as the input of the FBDD Transformer, and the inter-frame information is also considered to obtain four feature vectors of the person, background, head, and face to improve the detection efficiency and accuracy. FBDD distance is proposed to determine the type of video forgery. Constructing a rich dataset of video forgery methods containing face forgery and background forgery improves the accuracy of the model on different datasets. The FBDD algorithm was experimented on several commonly used and article-constructed datasets and has higher accuracy and stronger image degradation generalization compared to frontier detection algorithms.

Recently, there has been a growing interest in the field of computer vision and deep learning regarding a newly emerging problem known as action quality assessment (AQA). However, most researchers still rely on the traditional approach of using models from the video action recognition field. Unfortunately, this approach overlooks crucial features in AQA, such as movement fluency and degree of completion. Alternatively, some researchers have employed the transformer paradigm to capture action details and overall action integrity, but the high computational cost associated with transformers makes them impractical for real-time tasks. Due to the diversity of action types, it is challenging to rely solely on a shared model for quality assessment of various types of actions. To address these issues, we propose a novel network structure for AQA, which is the first to integrate multi-model capabilities through a classification model. Specifically, we utilize a pre-trained I3D model equipped with a self-attention block for classification. This allows us to evaluate various categories of actions using just one model. Furthermore, we introduce self-attention mechanisms and multi-head attention into the traditional convolutional neural network. By systematically replacing the last few layers of the conventional convolutional network, our model gains a greater ability to sense the global coordination of different actions. We have verified the effectiveness of our approach on the AQA-7 dataset. In comparison to other popular models, our model achieves satisfactory performance while maintaining a low computational cost.

Currently, most of the bladder-type accumulator test benches used have problems such as large installed capacity, severe energy loss, and incomplete testing functions. This article focuses on the research of the testing requirements and power recovery issues of bladder-type accumulators. A new type of hydraulic schematic and electrical control system for power recovery bladder type accumulator testing based on mechanical compensation with dual output shaft motor transmission, differential charging and discharging of dual accumulators, and parallel addition of oil replenishing accumulators has been designed. A measurement and control software based on the LabVIEW platform has been developed. The experimental results show that the system's power recovery efficiency can reach 60.5% by adopting a scheme of mechanical compensation type power recovery, differential charging and discharging of energy accumulator group, and parallel addition of oil replenishing energy accumulator. The testing system can accurately, efficiently, and conveniently complete various performance tests of hydraulic bladder accumulators.

The location of a Tunnel Boring Machine (TBM) is prone to external environmental and internal system noise interference. To address TBM vibration characteristics, this study analyzed the sources of excavation data noise, the need for noise reduction, and the methods used, using excavation speed as the prime example. We validated the results through three actual tunnel projects. Our research showed that excavation parameters automatically collected by the TBM, such as excavation speed, cutter head thrust, and cutter head torque, contain noise and require filtering and noise reduction before data mining. Different cutoff frequencies correspond to varying filtering effects, and after considering both smoothness and accuracy, we recommend setting the Butterworth filter's cutoff frequency to 0.1.

In this paper, the performance analysis, strength simulation, and vibration test of the structure of the hot-melt bonding machine are carried out. This paper firstly analyzes the structure and performance of the new hot-melt adhesive machine from the functional analysis of the product, then uses the knowledge of mechanical design to analyze the strength and stability of the whole machine, and uses Solid Works to build a three-dimensional model and the simulation software Ansys Workbench. The strength simulation and evaluation of the model were carried out. Finally, the samples were made and vibration tests were carried out. The hot melt adhesive machine has passed the transportation simulation vibration test and met the functional requirements of transportation stability. The simulation and testing methods in this paper provide a technical reference and solution for engineers.

In the process of digital transformation of rail transit vehicles, the digitalization of production line equipment is an important content. Starting from the whole life cycle business of production line equipment, this paper improves the existing production line equipment index system, proposes the production line equipment data acquisition model, describes the equipment portrait, proposes the comprehensive evaluation system based on the equipment index model, and further explores the digital twin architecture of production line equipment based on the integration of multiple models. Improve the digital level of rail transit vehicle manufacturing production line equipment.

This paper presented a robust angle-only guidance and navigation algorithm for asteroid defense missions based on meta-reinforcement learning. A recurrent neural network, trained via proximal policy optimization, is used to map the line-of-sight angles captured in real-time by the onboard camera to the optimal thrust. The neural network effectively replaces the roles of the navigation and guidance system while simultaneously removing the dependence on dynamic and observation models. The guidance and navigation model is tested on numerical simulations of a simulated mission directed to asteroid Bennu. The objective is to enable the spacecraft to hit the asteroid precisely, despite the presence of scattered initial conditions, uncertain model parameters, thruster control error, and attitude control and measurement error.

Peg-in-hole assembly is one of the more typical tasks in the machining industry. Although robotic peg-in-hole assembly technology has been widely used in practical production, achieving efficient and reliable flexible peg-in-hole assembly by robots still poses significant challenges. The analysis of robotic flexible peg-in-hole assembly from a control point of view is presented. Firstly, the process of robot flexible peg-in-hole assembly is introduced. Then the application of traditional model-based assembly control is described, on the basis of which learning-based intelligent assembly control is discussed. The combination of intelligent methods and traditional methods will become an important future development trend, injecting new vitality into robotic flexible peg-in-hole assembly.

Projection-based multimodal 3D semantic segmentation methods suffer from information loss during the point cloud projection process. This issue becomes more prominent for small objects. Moreover, the alignment of sparse target features with the corresponding object features in the camera image during the fusion process is inaccurate, leading to low segmentation accuracy for small objects. Therefore, we propose an attention-based multimodal feature alignment and fusion network module. This module aggregates features in spatial directions and generates attention matrices. Through this transformation, the module could capture remote dependencies of features in one spatial direction. This helps our network precisely locate objects and establish relationships between similar features. It enables the adaptive alignment of sparse target features with the corresponding object features in the camera image, resulting in a better fusion of the two modalities. We validate our method on the nuScenes-lidar seg dataset. Our CAFNet achieves an improvement in segmentation accuracy for small objects with fewer points compared to the baseline network, such as bicycles (6% improvement), pedestrians (2.1% improvement), and traffic cones (0.9% improvement).

In order to study the effect of friction wear behavior on electrical signal transmission, a carrier friction wear test machine was developed which can be analyzed under different test conditions (load, speed and current). The testing machine builds a measurement and control system through LabVIEW platform to realize real-time acquisition and synchronous display of electrical signals, friction coefficient and temperature, and control of load size, and uses wavelet decomposition and arithmetic average to process data and PID control algorithm to revise load in real time. The experimental results show that the stability of the test machine and the accuracy of control system meet the test requirements.

Hydraulic cylinder replacement robot as a new type of engineering machinery has been increasingly used, but its end effector encounters vibrations in the process of clamping the object, so the accuracy of disassembling and assembling the cylinder will be reduced, thus reducing the replacement efficiency and affecting the user's experience. To address this problem, virtual prototyping technology is used to study the cylinder disassembly process under real working conditions. We use the 3D modeling software Solidworks to construct a model of the cylinder replacement robot. After that, kinematic analysis of the model is carried out, then a dynamics model is built in multi-body dynamics simulation software ADAMS to simulate the process of the robot grasping the object, as a consequence, the trajectory of the end effector is calculated. A controlled dynamic model is established with Simulink and Adams by using the co-simulation technique, and optimization is carried out by using the model. Results show that the optimized control parameter can effectively reduce the end effector vibration and improve the stability and accuracy of the work.

The centrifugal pump plays a key role in the ship water supply system, and cavitation is a common fault mode that leads to poor efficiency of centrifugal pumps. To select valid features as the criteria of cavitation fault diagnosis, we calculate RMS, peak factor, kurtosis factor, wave factor of raw data, and the data after empirical mode decomposition (EMD). Although the raw features can be obtained, they were still in a high-dimensional space and contain a lot of redundant information. So, Principal Component Analysis (PCA) method was used to decrease dimensions and extract sensitive features. As a case study, the data were obtained from a centrifugal pump fault simulation bench, and a variety of cavitation states were observed in the experiment. And a six-dimensional sensitive feature vector was determined through data analysis.

Perception function, as an important part of autonomous driving, ensures the safety and intelligence of driving. We collect the surrounding environment of driving through hardware sensors, providing a basis for subsequent decision-making of autonomous driving. In recent years, deep learning has made breakthrough progress in object detection. Based on this, this paper uses lidar point cloud data, combined with deep learning theory and method, to carry out three-dimensional target detection tasks around lidar point cloud data, and carries out theoretical analysis, method verification, and result analysis. A 3D object detection method based on LiDAR point cloud data is proposed. The backbone network of the network model corresponding to the method is VoxelNet's backbone network. After outputting the feature matrix, sparse point clouds are supplemented with point clouds, and then the feature matrix is decoded to generate candidate boxes. The model used in this paper on the KITTI data set can effectively solve the problem of 3D target detection in the process of autonomous driving and has good performance.