Rice is a plant with rounded hollow articulated culms, flat, well-attached leaf blades, and terminal spikes. Its cultivation and consumption shape the culture, diet, and economy of different groups, especially in Asia. However, farmers suffer great financial losses each year due to rice disease. Therefore, the identification and classification of rice diseases are very important. Prompt, timely, and accurate disease diagnosis prevents product loss and improves crop quality. This study focuses on the classification of whether rice paddy leaf is normal or has a disease (one of the following: bacterial leaf blight (BLB), bacterial leaf streaks (BLS), bacterial panicle blight (BPB): heart, downy mildew, hispa, and rice tungro disease (RTD)) using deep learning-based algorithms such as EfficientNet-b0, MobileNet-v2, and Places365-GoogLeNet. The best model for this simulation was found to be EfficientNet-b0 with an average accuracy of 97.74%.

Since magnetic field is penetrating and harmless to human body, magnetic soft robots driven by magnetic field have great potential in medical fields. Thus, magnetic soft robots have attracted wide attention. However, the current researches mainly focus on the design of a single magnetic soft robot. Multiple magnetic soft robots also deserve to be studied due to their applications in collaborative operation. This paper presents a new design and fabrication method of multiple magnetic small-scale soft robots with different magnetic strength, size, and length-width ratio. The robots can be controlled to move in different motion modes and motion states under identical magnetic field. By analyzing their magnetic response property, which is the switching conditions between the two motion modes, and analyzing their states of the walking motion, two robots are selected from a batch of fabricated robots to carry out experiment. The results show that the two robots can move in different motion modes in the identical magnetic field.

Traditional congestion-control algorithms based on packet loss are more favorable for short round-trip time (RTT) flows. However, when the flow occupies the bottleneck bandwidth, the effect of bottleneck bandwidth and round-trip propagation time (BBR) congestion-control algorithm is on contrary. A short RTT flow may become “starved” after multiple iterations, which leads to a significant fairness problem in the BBR. This study analyzed the convergence using the BBR algorithm principle and the effect of RTT on fairness. Thereafter, we proposed an improved algorithm termed BBR-f, constructed an inverse proportional function of RTT, and added a gain function to the congestion window (CWND) gain. A simulation experiment using the ns-3 network simulator platform shows that the algorithm can optimize fairness and improve bottleneck-linked bandwidth utilization. Finally, the Jain fair index was used to evaluate its effectiveness, and a comparison experiment with traditional algorithms was performed to verify the efficiency of the proposed method.

The rapid global spread of the coronavirus disease 2019 (COVID-19) is now a reality. China has taken urban traffic control measures to prevent and control the epidemic, but this has prevented the flow of people between cities. This study investigates the mechanism of the impact of urban traffic control measures on the intercity population flow in China using the one-way causal measurement method. The results show that the impact of urban traffic control measures on the intercity flow of the population changes with time. Based on this, this study makes scientific suggestions for the government on how to reasonably undertake traffic control measures.

The rotating inverted pendulum is a nonlinear, multivariate, strongly coupled unstable system, and studying it can effectively reflect many typical control problems. In this paper, a parameter self-tuning fuzzy controller is proposed to perform the balance control of a single rotating inverted pendulum. Particle swarm optimization is used to adjust its control parameters, and simulation experiments are performed to show that the system can achieve stability with the designed parametric self-tuning fuzzy controller, with control performance better than that of the conventional fuzzy controller. Furthermore, the leader-follower control strategy is used to realize the cooperative control of multiple rotating inverted pendulums. Two QUBE-Servo 2 rotating inverted pendulums are used for a cooperative pendulum swing-up experiment and stabilization experiment, and the effectiveness of the proposed cooperative control strategy is verified.

Tremors are a symptom of several disorders of the central and peripheral nervous systems, such as Parkinson’s disease (PD). Exoskeletons have been investigated as noninvasive tremor suppression alternatives to medication and surgery. The challenge in musculoskeletal tremor suppression is attenuation of tremor motion without impeding the patient’s voluntary motion. Linear low-pass filters (LPFs) are commonly used for tremor removal. This study presents an alternative method based on an adaptive filter. We compare the effectiveness of the LPFs and adaptive filter on the recorded acceleration signals from an online database.

This study presents the bifurcation analysis and stabilization via the impulsive control of a fractional-order gene regulatory network with time delay. First, the author chooses the transcription rate k as the bifurcation parameter and obtains the Hopf bifurcation condition by analyzing its characteristic equation. The research shows that Hopf bifurcation occurs when the transcription rate k exceeds a critical value. This bifurcation behavior may destabilize the system. Subsequently, the author designs an impulsive controller to stabilize the system. Finally, simulation examples are used to verify our theory.

Self-attention has recently emerged as a technique for capturing non-local contexts in robot vision. This study introduced a self-attention mechanism into an intersection recognition system to capture non-local contexts behind the scenes. This mechanism is effective in intersection classification because most parts of the local pattern (e.g., road edges, buildings, and sky) are similar; thus, the use of a non-local context (e.g., the angle between two diagonal corners around an intersection) would be effective. This study makes three major contributions to existing literature. First, we proposed a self-attention-based approach for intersection classification. Second, we integrated the self-attention-based classifier into a unified intersection classification framework to improve the overall recognition performance. Finally, experiments using the public KITTI dataset showed that the proposed self-attention-based system outperforms conventional recognition based on local patterns and recognition based on convolution operations.

In this study, we applied natural language processing (NLP) techniques to texts of excavation reports on buried cultural properties to calculate the degree of similarity between the reports for determining archaeological sites that have a high degree of similarity. Specifically, we validated whether the similarity of sentence embeddings in the excavation reports of these sites is consistent with the existing classification. Four archaeological sites classified in existing archaeological research papers were used. For validation, 128 excavation reports from the four sites were used; sentence embeddings were obtained using Doc2Vec. We obtained the following results: 1) In applying NLP to excavation reports for determining the similarities of archaeological sites, merging the texts for each site into a single document and then processing it was more preferable than processing it in separate volumes of the excavation report. 2) The similarity based on sentence embedding of excavation reports using Doc2Vec was more consistent with the classification of the characteristics of archaeological sites than term frequency–inverse document frequency (TF-IDF). 3) When targeting a specific period, the sentence embedding exclusively for the text of the relevant period is consistent with the classification of the characteristics of the archaeological site from the artifacts and structural remains of that specific period. 4) When a specific period is targeted, the exclusive sentence embeddings of that period, obtained through the additive compositionality of sentence embeddings, can be used to classify the characteristics of archaeological sites based on the artifacts and structural remains on that period. Consequently, the similarities of texts based on NLP can reflect the similarities of archaeological sites. This holds true even for excavation reports that include spelling inconsistencies, optical character reader misrecognition, and garbled words.

The arch structure of human foot absorbs impact and assists push-off movements during walking. The objective of this study is to introduce arch structures and toe joints into a biped robot, verify the effects, and devise walk control methods by dynamics simulation. We simulated the upright state and start of walking using Choreonoid. The results confirmed that the arch structure improved the impact absorption and stability in the anteroposterior direction. In addition, the arch structure could be expected to smooth the load transfer between the supporting legs during the step change.

To improve the space utilization rates of intelligent stereo garages, their key execution equipment—parking robots—must be highly flexible; therefore, the chassis of parking robots use steering wheel systems. However, this design increases the complexity of robot motion control and path-tracking systems. Because of the large sizes of parking robots, large position and angle deviations can lead to derailment events, posing significant risk to users. Thus, in this study, based on the characteristics required for parking robots, a diagonally arranged parking robot with double steering wheels was developed and its kinematic model was established using the velocity–geometry method. For the positioning of the robot in a garage, a fusion positioning algorithm based on quick response (QR) codes and track calculation was used, considering the requirements of real-time response and accuracy. A discrete proportional-integral-derivative (PID) path-tracking algorithm for the synchronous compensation of the position and angle deviations was also devised. Using the Visual C++ platform, a path-tracking system that uses an industrial computer and motion control card was developed. Experiments on the path-tracking capability of the parking robot showed that, using the discrete PID path-tracking algorithm, it can track a planned path well. On Z- and U-type paths, its maximum tracking deviations were reduced by 84.8% and 64.0%, compared with the unused path-tracking system, reaching 10.4 mm and 10.9 mm, respectively. Thus, the tracking accuracy was significantly improved, proving that the developed robot well satisfies the path-tracking requirements of parking robots.

Causal relation extraction is a crucial task in natural language processing. Current extraction methods have problems, including low accuracy of causal-event division and incorrect extraction of important semantic features. This study uses the bidirectional long short-term memory (BiLSTM) and attentive convolutional neural network (ACNN) models to construct a cascaded causal relationship extraction model to improve the precision of the extraction. The model uses two kinds of labels and then divides the causal event boundary after determining the relationship between the front and rear causal events. It automatically learns semantic features from sentences, reducing the dependence on external knowledge and improving the precision of extraction. The experimental results demonstrate that the precision of causality extraction can reach 81.67% and the F1 value can reach 83.2%.

Recently, intelligent city construction has been promoted with the development of the Internet of things (IoT). The edge IoT gateway plays a critical role as the data aggregation core and processing center. Most existing gateways mainly solve heavy data storage and processing loads in cloud computing centers. There is less attention paid to multi-protocol data transmission and fusion. However, multiple products with different protocols in an IoT system require a flexible gateway compatible with multiple protocols. This paper proposes a multi-protocol edge gateway. The frame design was based on the actual demand for edge data acquisition. The gateway hardware platform used an RK3399 chip transplanted from the embedded operating system. It could support simultaneous multi-protocol access to ZigBee, LoRa, Bluetooth, and Wi-Fi. We combined the plug-and-play (PnP) hardware device access detection scheme with the system onboard interface driver to realize dynamic access detection and unified device management. In addition, the gateway also integrated data storage and access functions and partial edge computing functions. Finally, the experiment results verified that the multi-protocol edge gateway could meet the demand for data access and device control.

Currently, Japan’s Road Traffic Act uses the breath alcohol concentration (BrAC), which is the concentration of alcohol in 1 L of breath, to enforce the law. However, 400 drunk driving accidents occurred in 2020 when the BrAC was below 0.15 mg/L, which is legal limit in Japan, and cognitive function may be impaired at this concentration. Therefore, we investigated the effects of alcohol consumption at levels below a BrAC of 0.15 mg/L on cognitive function and its change over time using P300, a type of event-related potential, and the concomitant negative variation (CNV). Beer with a 5% alcohol content was used as the drinking load, and the amount of alcohol consumed was determined using a derivation formula that varied according to the subject’s body weight. Electroencephalogram (EEG) measurements were obtained five times before drinking and 10, 30, 50, and 70 min after, and participants were given a go/no-go task to induce event-related potentials during the measurements. The BrAC was measured immediately before the EEG measurement was performed. The results revealed that the alcohol concentration in the exhaled breath increased significantly (p<0.05) at 10, 30, and 50 min after drinking compared with immediately before. In addition, the P300 latency was significantly prolonged (p<0.05), and the early and late CNV amplitudes were significantly increased (p<0.05) in the event-related potentials during these time periods. Moreover, no significant difference was observed in expiratory alcohol concentration at 70 min post-drinking, but significant changes in event-related potentials still occurred. Therefore, our results suggest that cognitive function is impaired in patients with low levels of alcohol consumption, when the BrAC is below the legal limit, and even when the BrAC recovers after the end of alcohol consumption.

Human motion forecasting is a necessary variable to analyze human motion concerning the safety system of the autonomous system that could be used in many applications, such as in auto-driving vehicles, auto-pilot logistics delivery, and gait analysis in the medical field. At the same time, many types of research have been conducted on 3D human motion prediction for short-term and long-term goals. This paper proposes human motion forecasting in the 2D plane as a reliable alternative in motion capture of the RGB camera attached to the devices. We proposed a method, the time series self-attention approach to generate the next future human motion in the short-term of 400 milliseconds and long-term of 1,000 milliseconds, resulting that the model could predict human motion with a slight error of 23.51 pixels for short-term prediction and 10.3 pixels for long-term prediction on average compared to the ground truth in the quantitative and qualitative evaluation. Our method outperformed the LSTM and GRU models on the Human3.6M dataset based on the MPJPE and MPJVE metrics. The average loss of correct key points varied based on the tolerance value. Our method performed better within the 50 pixels tolerance. In addition, our method is tested by images without key point annotations using OpenPose as the pose estimation method. Resulting, our method could predict well the position of the human but could not predict well for the human body pose. This research is a new baseline for the 2D human motion prediction using the Human3.6M dataset.

Substation equipment faults are typically related to the heating of equipment components. The hotline tap clamps of substation are critical components for carrying load currents and thermal fault potential. As a result, a new hybrid early warning approach for hotline tap clamp faults in substation equipment is presented. A two-dimensional Otsu algorithm is used to coarse-segment infrared images to minimize the subsequent complexity. Since the Chan–Vese (CV) model is insufficiently accurate for image segmentation with uneven grayscale, then the differential data obtained by the Prewitt operator to identify the goal edges are combined with the CV model to improve segmentation accuracy. The improved CV model achieves excellent segmentation of the hotline tap clamp in the substation. The temperature statistics are utilized for the segmented images, and the hotline tap clamp fault warning is realized based totally on the relative temperature difference. Finally, the experiments exhibit that the method can enhance the segmentation impact of infrared images and obtain the goal of fault warning.

Logistics tracking technology at normal temperature is quite mature, but there are few tracking methods for the high-temperature production process. The main difficulties are that the label materials generally used cannot withstand the high temperature for a long time, and the detection devices are vulnerable to environmental impact. A high-temperature logistics tracking solution was developed for a carbon anode used in an aluminum electrolysis factory. It is based on concave coding and a multiscale low-level feature fusion and attention-DeepLabV3+ (MLFFA-DeepLabV3+) network extraction technique for the coded region of the concave coding. The concave coding is printed on the product as a tag that can endure a high temperature of more than 1,200°C, ensuring its integrity and identifiability. Because there is no obvious color distinction between the coding area and the background, direct recognition is ineffective. The MLFFA-DeepLabV3+ network extracts the coding region to improve the recognition rate. The DeepLabV3+ network is improved by replacing the backbone network and adding of a multiscale low-level feature fusion module and convolutional block attention module. Experimental results showed that the mean pixel accuracy and mean intersection over union of the MLFFA-DeepLabV3+ network increased by 2.37% and 2.45%, respectively, compared with the original DeepLabV3+ network. The network structure has only 11.24% of the number of parameters in the original structure. The solution is feasible and provides a basis for high-temperature logistics tracking technology in intelligent manufacturing.

China is one of the largest wheat production countries in the world. The wheat quality determines the price and many other aspects. The detection methods of wheat quality mainly depend on manual labor. It costs high amount of manpower and time, and the classification results are partly affected by different individuals. With the development of machine vision, an automatic classification system was presented in this study. A wheat unsound kernel identification method based on the improved YOLOv5 algorithm was designed by adding efficient channel attention (ECA). Compared with convolutional block attention module (CBAM) and squeeze-and-excitation network (SENet), the improved YOLOv5 algorithm was selected to fit the model better. The recognition results showed that YOLOv5 with the addition of the attention mechanism had a significant improvement in average accuracy over that without it. The most significant improvement was observed with the addition of ECA-YOLOv5, with an average accuracy of 96.24%, a 10% improvement over the other two models, and a 13% improvement over the original YOLOv5. This satisfied the application requirements for detection of wheat unsound kernel.

The differentiable neural computer (DNC) is a neural network model with an addressable external memory that can solve algorithmic and question-answering tasks. Improved versions of the DNC have been proposed, including the robust and scalable DNC (rsDNC) and DNC-deallocation-masking-sharpness (DNC-DMS). However, integrating structured knowledge and calculations into these DNC models remains a challenging research question. In this study, we incorporate an architecture for knowledge and calculations into the DNC, rsDNC, and DNC-DMS to improve their abilities to generate correct answers for questions with multi-hop reasoning and provide calculations over structured knowledge. Our improved rsDNC model achieves the best performance for the mean top-1 accuracy, and our improved DNC-DMS model scores the highest for the top-10 accuracy in the GEO dataset. In addition, our improved rsDNC model outperforms other models in regards to the mean top-1 accuracy and mean top-10 accuracy in the augmented GEO dataset.

In a coil magnetometer, the size and uniformity of the bias magnetic field generated by the Helmholtz coil directly determine the accuracy of the solution of the geomagnetic direction. The design of traditional spherical coils relies heavily on the manual experience or mathematical derivation, making it difficult to obtain optimal parameters or requiring larger spherical coils. To address the problem, first, a coaxial symmetrical spherical coil model that improves space utilization was established. Second, an optimal design method for the spherical magnetic field generation coil based on a differential evolution algorithm was proposed. Third, the optimal bias magnetic field was obtained without increasing the volume of the coil. The verification results showed that the magnetic non-uniformity and magnetic gradient of the bias field generated by the optimized coil were reduced by 63.2% and 82.8%, respectively.

This paper develops an intelligent electrode conversion device for distributed high-density resistivity meter. The direct-current (DC) power line communication (PLC) technology is used for developing the instrument. The relay switch array based on a three-wire cable is designed and circuit simulation verification is carried out. The results demonstrate the feasibility of the proposal.

In collaborative control of multi motors for heavy-duty locomotives, the output value of the motor frequently exceeds its maximum allowable value during power redistribution. This results in a saturated motor owing to the power redundancy of each wheel set of the train. In this study, an algorithm for the tracking control of a consistent total amount of the extended observer through anti-saturation is proposed. First, mathematical models of multi-motor traction systems are developed. The system includes uncertain parameter perturbations and external perturbations. Second, a new type of extended sliding mode observer (ESMO) is designed to reduce the influence of the tracking effect on the input saturation of the system. Subsequently, for collaborative control of multi motors in heavy-duty locomotives, a new scheme for dynamic and auxiliary anti-saturation compensation (anti-windup) is established. The perturbation observation results and the systems’ auxiliary status are respectively input into the sliding mode controller (SMC). A traction total-amount coordinated tracking control (TACTC) of multi motors is achieved to ensure consistency of the system’s total output torque and the given traction characteristic curve. Finally, simulations and tests are performed on the motor actuators to demonstrate a good control effect.

This paper compares Japanese and multilingual language models (LMs) in a Japanese pronoun reference resolution task to determine the factors of LMs that contribute to Japanese pronoun resolution. Specifically, we tackle the Japanese Winograd schema challenge task (WSC task), which is a well-known pronoun reference resolution task. The Japanese WSC task requires inter-sentential analysis, which is more challenging to solve than intra-sentential analysis. A previous study evaluated pre-trained multilingual LMs in terms of training language on the target WSC task, including Japanese. However, the study did not perform pre-trained LM-wise evaluations, focusing on the training language-wise evaluations with a multilingual WSC task. Furthermore, it did not investigate the effectiveness of factors (e.g., model size, learning settings in the pre-training phase, or multilingualism) to improve the performance. In our study, we compare the performance of inter-sentential analysis on the Japanese WSC task for several pre-trained LMs, including multilingual ones. Our results confirm that XLM, a pre-trained LM on multiple languages, performs the best among all considered LMs, which we attribute to the amount of data in the pre-training phase.

The importance of information-technology (IT)-enabled interactions in promoting interfirm trust has been documented in literature. However, the currently available findings are limited because the role of IT-enabled informal interactions has been overlooked and the categorization of trust is unclear. To obtain detailed knowledge about the relation between IT-enabled interactions and trust, this study categorizes not only the IT-enabled interactions into formal and informal types but also trust into goodwill and competence trust. Moreover, this study explores the moderating role of relationship duration on the aforementioned relationships. Through a survey of 240 firms in China, this study made the following observations: (1) IT-enabled formal and informal interactions can promote both competence and goodwill trust; (2) compared with IT-enabled informal interactions, IT-enabled formal interactions promote competence trust, and a comparable extent of goodwill trust; (3) relationship duration strengthens the positive link between IT-enabled formal interactions and competence trust; however, it has no moderating impact on the link between IT-enabled formal interactions and goodwill trust; (4) relationship duration weakens the impact of IT-enabled informal interactions on goodwill trust; however, it has no significant moderating role on the relation between IT-enabled informal interactions and competence trust. Overall, this study enriches the knowledge regarding the relationship between IT-enabled interactions and interfirm trust.