To achieve ultra-high reliability and ultra-low latency required for Beyond 5G or 6G, low error rate decoding method is needed with short code lengths and high code rates. Additionally, a universal decoding method that can be applied to any code which appears in the future be desirable. The maximum likelihood decoding is the best decoding method in information theory, but it is considered difficult to implement due to its huge computational requirements. Recently, Soft GRAND (SGRAND) has been proposed as an effective decoding method for any high-rate, short code length block code in any continuous channel. This method obtains the same decoding result as the maximum likelihood decoding by checking the most likely noise sequantially. However, SGRAND has a problem in which the size of the noise candidate list held internally increases linearly with each noise check. This paper proposes a method to reduce the size of the candidate list in SGRAND, while maintaining the same decoding results as maximum likelihood decoding, by modifying the noise candidate generation algorithm in SGRAND.

Quantum illumination has been a hot topic in quantum sensing since first proposed by S. Lloyd in 2008. Quantum illumination is an entanglement-based protocol that is able to detect the existence of a target in an environment with low-energy transmissivity and strong thermal noise. The Bell state and the two-mode squeezed vacuum (TMSV) state as an entangled light source have been widely discussed. In our previous study, we clarified that quantum illumination using the quasi-Bell state offers good performance at error probability even though easing the above-mentioned restrictions on the environment. In this study, we considered and discussed a physical implementation of a measurement scheme for the quasi-Bell state-based quantum illumination using a beam splitter and photon counters. As the result, although the measurement scheme is not an optimum scheme, it still offers a clear advantage over that using the TMSV state, which allows optimum quantum detection, under almost situations.

Some stroke patients are unable to open their paralyzed hand with their own intention due to spasticity or muscle weakness. In our previous work, we developed a finger extensor facilitation training device “iPARKO” that imitates finger extensor facilitation technique. In this paper, we developed a hand fixation brace of the iPARKO for stroke hemiplegics and conducted the experiments to verify the immediate effect of finger extension for six stroke survivors. We compared the effects of iPARKO training and finger extensor facilitation technique. Comparison results showed that the amount of improvement in active range of motion with iPARKO training reached 48.3% more than that with finger extensor facilitation technique, and larger in the two participants. The results that the uniformity of each fingertip load in stroke survivors was equivalent to that of healthy participants confirmed the effectiveness of the new hand fixation brace. In addition, it was confirmed that Modified Ashworth Scale Score improved in all participants with iPARKO training and that it was effective in reducing spasticity.

We have developed a new loss function focusing on grasping product to automate the display of products by robots. When robots grasp product, it is required to select a grasp product from several products and detect it accurately. In recently years, object detection methods using deep neural network(DNN) has become more common. But even the latest DNN often does not recognize the boundaries of adjacent objects. So, the DNN erroneously detects two products as one product. We address the new problem of correctly detecting adjacent products. To solve the problem, we focus on the fact that the robot grasps only one product at a time, and have developed a new loss function called “mutual loss”. The mutual loss is larger when the DNN erroneously detects adjacent products as one product, so that the DNN correctly detects the products to be selected. Experiments on the OCID dataset showed that IoU of detected object and grasping success rate improved by 6.6pts and 5.8pts, respectively, when DNN uses our mutual loss.

In this study, we developed a new method to translate unknown products into known products while maintaining the pose of the products, in order to perform pose estimation of unknown products with zero-shot. Pose estimation methods using Neural Networks (NNs) are highly accurate, but their success rate of pose estimation is significantly decreased for unknown products. In retail stores, products are frequently changed, and it is time consuming to prepare data for each new product. Therefore, the purpose in this study is to improve the success rate of pose estimation with zero-shot. Since the success rate of pose estimation are high for known products, unknown products are translated into known products. However, even if only the appearance of the product is translated into the known product, the success rate of pose estimation cannot be improved because the pose of translated appearance often changes. Therefore, the coarse pose of the product is used as input for image translation NN to translate the product into a known product that maintains its pose. Experimental results showed that using pose-maintaining image translation improved the success rate for the two methods of pose estimation which are state-of-the-art on different datasets.

This study proposes a method for estimating court boundaries and displaying player positions and trajectories for broadcast handball videos. The videos typically cover only about one-third of the court, and the camera angle changes from left to right to follow the players. Initially, an object detection algorithm YOLOv5 is used to detect a goal object in the video, and the camera direction is determined from its position. Court boundaries are estimated in two ways: one using a bounding box around the goal object, and the other with a red area marked on the court. Once the court boundaries are estimated, DeepSort, an object tracking model, is employed to track the players, and their positions and movements are displayed on a top view of the court. The proposed methods were compared by conducting experiments on videos from different venues, and the results indicate that the estimation rates of the two methods were 80.7% and 85.3%, respectively.

Eech ballpark has a different size in baseball. It could be easily imagined that there would be many home runs in a small ballpark. Park Factors (PF) are introduced in baseball to quantify the effect of each ballpark on the results (e.g., home runs). In this paper, we assume that each plate appearance can be modeled as a match-up between a batter against a pitcher with a ballpark. The effects of each ballpark will be distilled by using a logistic regression method. Numerical verification shows that the proposed method performs better than conventional PF. The verification is based on the results of more than 140 thousands plate appearances from the 2017 to 2021 Japanese Pacific League Baseball seasons.

In basketball, each player is evaluated based on measured statistics, e.g., points scored, field goal percentage, and so forth. Impact metrics are defined as those that attempt to quantify contribution of every player in a single value. In this paper, a novel impact metric named “value of contribution (VC)” is defined and validated. This metric is based on the win probability model and pairwise comparison between two lineups, i.e., two sets of five players on the court, of competing two teams. In construction of the win probability model, we developed a model that guarantees monotonicity on the score difference. For every play, the lineups are rated based on the change on win probability. VC of a player is defined by the weighted average of the ratings of lineups that include the player (the weight is playing time). By definition VC of a player estimates the win probability change per unit time when he is on the court. The proposed VC is calculated for the players in Japanese B. LEAGUE, and several case studies will inform how to use this metric in the analysis of squads.

Outdoor play is important for children because it is fun, improves communication, and provides adequate exercise. In addition, the environment and conditions for traditional play have changed due to the coronavirus pandemic, making it necessary to consider new rules of play. In this study, we focus on tag, one of the games, and consider rules that can reduce the differences in physical advantage and disadvantage due to foot speed. In this study, we consider three rules. (1) the “number selection behavior” rule (R1-1), in which the chaser actively pursues the runner who has been the chaser less frequently; (2) the “replacement condition A” (R2-1), in which the replacement condition between the chaser and the runner is multiple touches; and (3) the “replacement condition B” (R2-2), in which the replacement condition between the chaser and the runner is one touch and a rock-paper-scissors victory by the chaser. We analyze the effects of these three conditions on the game of tag using multi-agent simulation.

In order to compensate non-linearity of any electronic devices, we have proposed small scale exponentiation conversion circuit by utilizing subthreshold operation of the MOSFET up to now. The exponentiation conversion circuit multiplies the logarithmically converted input signal by the power exponent value to perform exponential conversion. The power supply voltage of the power conversion IC has been reduced to 3.3V, however, the characteristics of the MOSFETs have changed with the lower voltage. Also, since the performance of the exponentiation conversion IC is determined by the exponential and logarithmic conversion accuracy of the signal in subthreshold operation, the current density and bias conditions of the MOSFET must be carefully designed. In this paper, we report the details of the MOSFETs’ optimum operating conditions, which were established by detailed analysis of the signal dynamic range, subthreshold leak current and subthreshold swing in the subthreshold region of the MOSFETs when the source voltage, supply voltage and Temperature are varied.

The analysis of knee motion during walking is essential for understanding the mechanisms of knee joint contact force (KJCF), a factor associated with knee joint pain and related joint diseases. A comprehensive analysis of whole-body motion can provide valuable insights and practical strategies for mitigating KJCF. This study aimed to identify body segments whose motion is related to KJCF using a Convolutional Neural Network (CNN). We used a gait database to obtain three-dimensional motion data and calculated their KJCF using a musculoskeletal model. In addition, the peak values of KJCF during gait were classified into five classes to develop a learning model with CNN. Visualization using Gradient-weighted Class Activation Mapping revealed that the regions of interest identified by the trained CNN model were the bases of the middle and ring fingers of the right hand, the outside of the right thigh, the left side of the hip, and the thumb side of the left wrist. These findings suggest that the motions of the hand markers, which change with the arm swings, impact the variation of KJCF.

Conventionally, the adaptive body area for prism adaptation (PA) has been limited to the visible range of prism glasses, especially the unilateral upper limbs. In this study, we developed a virtual reality system that extends PA’s adaptive body range. This system sets the visible area of the target and introduces a symbolic representation of the body position to realize a head position adaptation task. As a result, the head position adaptation task generates PA not to individual body parts but to a collection of multiple motor units, such as the upper trunk. Furthermore, experimental results in healthy subjects showed that the head position adaptation task could generate PA in the entire upper trunk and displace it to the left anteriorly. Therefore, the head position adaptation task is expected to be effective for the postural correction of USN patients with upper trunk deviated to the right posteriorly. Moreover, the developed system has potential applications in a wide range of fields, such as detecting cerebellar disorders.

LED devices are becoming increasingly popular because they are more energy-efficient lighting devices than fluorescent ones. Currently, LED devices in IoT devices enable additional functionality, such as network and speaker functions. However, due to those functionalities, we encounter a new risk that LED devices may leak unintended signals, i.e., side-channel information. In this paper, we use audio signals as the side-channel information contained in LED lights. The reason for employing the audio signal is that it is not only a real threat but also suitable for investigating the frequency limits of leaked side-channel information. We use smartphone cameras to obtain the LED light and try to reconstruct the audio signal in the experiments. As a result, we show that information retrieval from the original audio is feasible. Furthermore, the image sensor's specification, such as shutter mechanism, frame rate, and blanking period, is highly related to the precision in acquiring the side-channel information.

The chaotic secure communication issue is addressed in this research using event-triggered extended high-gain observers. The communication system is based on two-channel communication, in which the transmitter system transmits the output of the chaotic system in one channel and the encrypted message that is obscured by the states of the chaotic system in the second channel. An event-triggering mechanism is introduced for each channel to decide the irregular time instants to transmit out the signal such that the resource-demanding continuous communication is avoided. In the receiver system, two event-triggered extended high-gain observers are proposed respectively to recover the chaotic system's states and unknown nonlinear function simultaneously, as well as the continuous encrypted message signal, from intermittently received signals. Based on the estimated signals, the message signal is decrypted by the inverse function of the encryption function. Theoretical analysis demostrates that Zeno phenomenon is not present and that the signal estimate errors diminish quickly to the area around the origin. Finally, simulation experiments are offered to demonstrate how well the proposed strategy works.

Automatic crack detection is an essential task for the effective maintenance of roads and structures. In recent years, deep learning has been widely applied to crack detection, and many models have been proposed. However, there isn't a standardized metric for the evaluation of crack detection methods nor reliable comparisons. This paper addresses this issue by examining the evaluation metrics suitable for cracks, selecting useful ones, and suggesting improvements. Additionally, a crack detection evaluation software was proposed and tested on the CRACK500 dataset, comparing FPHBN and DAUNet crack detection methods. The results confirmed the efficacy of the proposed metrics and the proposed software in helping researchers evaluate and compare different methods efficiently and effectively.

In recent years, digital marketing in the retail industry is merging “online”, which is sales through the Internet such as EC sites, and “offline”, which is direct sales through physical stores. On the other hand, the use of e-commerce sites is not yet sufficiently developed in the market of retail stores such as home electronics mass retailers, and there is room for expanding user range. The purpose of this study is to extract potential online customers from customers who have never purchased from an e-commerce site and to implement efficient customer targeting. The paper provides a method for extracting knowledge about offline customers who have similar characteristics to online customers by applying explainable AI (XAI) method for XGBoost trained on customer data. In this study, by applying the proposed method to Oricon customer satisfaction survey data, we detected the difference between potential customers and non-potential customers, and confirmed the effectiveness of the proposed behavior.

We propose a system that enables simulation of flood damage and evacuation training at arbitrary locations throughout Japan in a VR space, and provides evacuees with intuitive evacuation directions when evacuation is necessary at their location.

The VR space is constructed to approximate the real world based on location information and simulates flood damage. AR evacuation support determines the danger of the direction and alerts the user with color and sound.

Evacuation training was conducted using the VR space. As a result, subject’s awareness of disaster preparedness was improved. The color and sound evacuation support using AR technology freed up both hands and made it possible to show safe or dangerous directions more clearly than hazard maps.