In recent years, wireless sensing techniques, such as Wi-Fi and Ultra-Wideband (UWB) signals, have gained attention for activity recognition due to their ability to address privacy concerns associated with traditional computer vision methods. While UWB Channel Impulse Response (CIR) is believed to be prominent approach as well as Wi-Fi Channel State Information (CSI), research on its application in activity recognition remains limited. Previous studies have not fully explored the brevity of single measurements or the potential for feature extraction from CIRs. This paper presents a novel approach to robust device-free activity recognition by exploiting periodic UWB CIR samples. Utilizing multi-level wavelet packet decomposition (WPD) and a customized attention mechanism, the proposed method effectively combines multi-resolution features, improving recognition accuracy and reducing the need for extensive fine-tuning. Experiments conducted in various scenarios validate the performance of the proposed approach, with ablation studies demonstrating the superiority of multi-resolution analysis over short-time Fourier transform (STFT) and highlighting the cost efficiency of the method.

There is increasing growing demand for indoor altitude estimation using smartphone barometers. However, disturbances caused by ventilation fans pose a crucial challenge in barometric pressure–based altitude estimation methods. Thus, this paper proposes a method to detect disturbances caused by ventilation fans with only barometric information. Recognizing ventilation environments is challenging due to the influence of seasons and weather on barometric pressure values. The proposed method captures the characteristics of air fluctuations, weather conditions, and temporal information, and it uses machine learning techniques to identify the presence of disturbances caused by ventilation. Experimental results show that the proposed method outperforms conventional detection methods.

The demand for indoor altitude estimation using smartphone barometers is increasing. It is known that disturbances such as changes in temperature and ventilation systems affect altitude estimations using barometric values. In this paper, we propose a correction method to compensate for the disturbance caused by ventilation systems. The amount of depressurization in a room due to a ventilation system varies depending on the ventilation capacity and the room conditions, making it difficult to determine precisely when the ventilation system is operating based on the extent of depressurization. Our approach uses machine learning to detect only the transitions in the ventilation environment in a data-driven way and performs a correction by calculating the depressurization level from the preceding data at the time of detection. We conducted experiments in real-world environments that showed the proposed method is significantly superior in performance to previous correction methods.

Indoor position information, which is difficult to obtain by GPS, can be used for various services and applications, thus indoor localization methods have been widely researched. Among them, device-free indoor localization does not require the target persons to possess a localization device, such as a smartphone, which can support the localization of all persons in the environment. Many methods using RSSI and Wi-Fi CSI have been proposed as device-free indoor localization using radio waves. However, RSSI is easily affected by environmental factors, such as multipath propagation. Wi-Fi CSI also has the disadvantage that there is no standard, so it relies on special hardware and software. Therefore, we propose device-free multi-person indoor localization using ToF information, which is less susceptible to noise. ToF information in indoor localization has mainly been used for highly accurate estimation of the receiver position, while this paper proposes a new application. In addition, distance measurements using ToF of Wi-Fi and UWB have been standardized and can be implemented with commercial equipment, which is highly practical. In this research, we considered the change in the distance measurement due to radio wave occlusion by a human body and built a device-free indoor localization system that can estimate multi-person position even if the model is trained on the data of only one person.

Systems that utilize and manage predictive models have become increasingly significant in industry. In the services offered by Yahoo! JAPAN, once a predictive model is utilized for recommendations, it is thrown away. Such models could however be reused for expanding the coverage of other recommendations. Here, our goal is to construct recommendation systems that expand the coverage of recommendations by effectively utilizing models which would otherwise be discarded. Another goal is to deploy such a recommendation system on real services and make practical use of it. In this paper, we describe a recommendation system that achieves these two goals by overcoming the challenges facing its deployment on real services. Specifically, we developed an optimization method that alleviates the psychological barrier against using the recommendation system and clarified the performance of our method in making real recommendations. An offline test and a large-scale online test on making real recommendations showed that our method substantially expands the coverage of recommendations. As a highlight of the results, our method made recommendations to 76.9 times more users at the same level of recommendation performance as the currently used recommendation system by the service. Overall, the results show that our method has a huge impact on services and can be applied to real recommendations.

Precise estimation of heading, which is a key component of pedestrian dead reckoning for indoor positioning methods, is needed. We propose a new method that improves the accuracy of the heading estimation by processing data at a resolution finer than one step. First, the sensor data are divided into segments finer than a step by using the norm of the acceleration vectors. Then, features to predict the heading are automatically extracted from the segments by using a neural network based on long short term memory, instead of using the statistics of steps as features as is done conventionally. Our method is more accurate than the conventional one for walking in straight lines and walking on curved trajectories.

Demand for recognition of purchasing behavior has been increasing due to the expectation that it would be profitable in marketing. A method of recognizing purchase behavior involves classifying customer behaviors into three states: "moving", "searching" (for a product) and "observing" (i.e., examining the label, etc.) on the basis of acceleration and angular velocity data obtained from the sensor of a smartphone. The method can recognize a user’s purchasing behavior without relying on external infrastructure such as cameras or radio frequency identifiers (RFIDs). However, its recognition accuracy is low; it falls below 30 percent especially in the searching state. In this study, we focus on the fact that the number of steps per time (step rate) is different in each state. That is, the customer’s step rate becomes slower as states change from moving to observing, as it is harder to examine products while on the move. In addition, we propose a two-stage classifier based on the differences in walking motion. Through these features and the classifier, we aim to improve the accuracy of behavior recognition. For an evaluation, we asked subjects to freely walk and shop for products in a supermarket and took movies of them. We compared the classification accuracies of three types of feature: base features, step-rate features, and the total feature combining the first two types. Our evaluation indicated the accuracy score of the step-rate features improved by 12.7% and that of the total feature improved by 18.2% relative to using the base features. In addition, a two-step classifier using the total features improve the accuracy score by 2.3% compared with the one-stage classifiers. By further subdividing purchase behaviors into ones such as "crouching motion", we can further improve accuracy and also recognize purchase behaviors in more detail.

Many different services acquire feedback from users, and many predictive models of users interests have been made for recommending items by using data acquired from different services. Here, although we would be able to get deep insights about a user's interest by connecting predictive models, interactions among services and potential synergies of their predictive models have been neglected. Moreover, in the era of big data, it should be relatively easy to develop a recommendation system that works across services by connecting predictive models and using existing technologies. However, the effect and conditions under which such the recommendation system works effectively have not been analyzed in detail.We have developed ModelRecycling FrameWork (MRFW) as an implementation of our idea of connecting predictive models to make recommendations across services. We experimentally evaluated the performance of MRFW by using an online questionnaire survey. The results show that MRFW significantly outperforms predictive models that make recommendations without considering the interaction between services or synergies of predictive models. We also clarified the conditions under which MRFW works effectively by comparing its performance with that of the individual models on which it is based in order to make use of the results of analysis for designing the method that connects predictive models.

There is growing interest in using augmented reality technology for gaming, navigation, and remote communication. Although 3D space models can be made manually or digitalized using specialized and expensive sensing devices like LiDAR (Light Detection and Ranging), which are costly and time-consuming, the recently developed ARCore for Android and ARkit for iOS are convenient and quick means of developing AR applications. Their recognition performance is poor for flat and monotone walls. This drawback is significant because most walls in indoor environments are flat and only a small portion of them are colorful enough to detect feature points.

This paper presents a resource manager to achieve scalable performance in Robot Operating System (ROS) for distributed environments. In robotics, using ROS in distributed environments via multiple host machines is trending for large-scale data processing, for example, cloud/edge computing and the data communication of point clouds and images in dynamic map composition. However, ROS is unable to manage the resources (e.g., the CPUs, memory, and disks) on each host machine. Therefore, it is difficult to use distributed environmental resources efficiently and achieve scalable performance. This paper proposes a resource management mechanism for ROS distributed environments using a master-slave model to execute ROS processes efficiently and smoothly. We manage the resource usage of each host machine and construct a mechanism to adaptively distribute the load to be balanced. Evaluations show that scalable performance can be achieved in ROS distributed environments comprising ten host machines using a real application (SLAM: simultaneous localization and mapping) processing large-scale point cloud data.

Wi-Fi fingerprint-based localization is known to be prominent for indoor positioning technology; however, it is still challenging on sustainability of its performance for long-term use due to distribution drifts of the signal strength across time. Therefore, the laborious continual surveys on fingerprint are inevitable. In this paper, we propose a new scheme for solving the large cost of maintaining common Wi-Fi fingerprint-based localization with machine-learning-based way by efficient incremental learning (retraining). Specifically, we propose a brand new retraining method, called GroupWi-Lo, that focuses on minimization of parameter variation with respect to the incremental surveys on fingerprint (i.e., calibration). Our method tries to keep the parameters of the previously trained model unchanged while minimizing the error on the dataset obtained in the last surveys. This formulation is helpful to keep robustness against overfitting from the limited size of the dataset per survey. The experimental results both in the lab and the uncontrolled environment show that GroupWi-Lo achieves competitive performance among the state-of-the-art methods, while its computational cost retains independent of the number of surveys compared with existing the semi-supervised approach and standard incremental training approach.

At present, Wi-Fi localization is the main approach to estimating location indoors. However, age deterioration of the localization model due to dynamic environmental changes degrades its accuracy. Therefore, periodic model recalibration is inescapable. Existing methods for doing this use transfer learning and a small set of additional and supervised datasets. However, the reference points to obtain these datasets are determined either randomly or comprehensively. Such poor datasets catastrophically destabilize model recovery after recalibration because overfitting occurs. We propose a new approach that detects anomalous reference points to gain felicitous supervised datasets in order to prevent overfitting. Unsupervised datasets obtained from off-the-shelf mobile navigation applications, i.e., user logs uploaded from phones, are used. Our approach is implemented in a system we call "No-Sweat Detective". The results of an experiment in a controlled environment demonstrate that No-Sweat Detective can detect anomalies caused by environmental changes, and the results of a five-month experiment show that No-Sweat Detective has redundancy against a complex open-space environment in the real world. In addition, it could suppress model age deterioration by up to 10.9% compared to existing methods.

This paper presents GPUhd, a graphics processing unit (GPU) resource management approach that combines Hadoop and a GPU to obtain scale-out and scale-up functionality. There are several researches that combine Hadoop and GPU. However, there are no researches that can schedule tasks in consideration of GPU resource on Hadoop. Moreover, these researches cannot use multiple distributed frameworks. GPUhd extends the Yet Another Resource Negotiator (YARN) management mechanism and distributed processing frameworks for the coordinated use of GPU resources in Hadoop. We extend the YARN scheduling algorithm to consider GPU resources and incorporate a resources monitoring function. GPU resources can be managed on the basis of existing development methods because GPUhd simply handles GPU resources as host memory and CPU resources. In addition, GPUhd achieves high-speed processing, e.g., the computational time required to calculate 2048 x 2048 matrix multiplication is approximately 25 times less than that required when using only a CPU with Hadoop. GPUhd achieves high scalability and excellent response times in a heterogeneous distributed environment.

Robot Operating System (ROS) promotes fault isolation, faster development, modularity, and core reusability and is therefore widely studied and used as the de facto standard for autonomous driving systems. Graphics processing units (GPUs) also facilitate high-performance computing and are therefore used for autonomous driving. As the requirements for real-time processing increase, methods for satisfying real-time constraints for ROS and GPUs are being developed. Unfortunately, scheduling algorithms specifying ROS's transportation (publish/subscribe) model, which can have execution order restrictions, are not being investigated, leading to the introduction of waiting time and degrading the responsiveness of the entire system. Furthermore, GPU tasks on ROS are also affected by the ROS transportation model, because central processing unit (CPU) time is occupied when GPU functions are launched. This paper proposes a loadable kernel module framework, called real-time ROS extension on transparent CPU/GPU coordination mechanism (ROSCH-G), for scheduling ROS in a heterogeneous environment without modifying the OS kernel and device drivers and then evaluates it experimentally. ROSCH-G provides a scheduling algorithm that considers ROS's execution order restrictions and a CPU/GPU coordination mechanism. Experimental results demonstrate that the proposed algorithm reduces the deadline miss rate and, compared with previous studies, makes effective use of the benefits of parallel processing. In addition, the results for the coordination mechanism demonstrate that ROSCH-G can schedule multiple GPU applications successfully.

Along with the downsizing of computers and sensors, user situation and state can be recognized by sensing, signal processing, and machine learning technologies in many places. In home environments, it is reported that the number of drowned people in a bathtub has increased to 4,886 in 2004, which is 1.7 times from 2004. In order to prevent the occurrence of such a fatal accident, it is important to grasp the state of people while bathing. In this study, we propose a method to identify people in the bathtub from changes in water level during bathing with a water pressure sensor installed at the bottom of the bathtub. In the evaluation experiments, the average identification accuracy for four subjects is as low as 65% in the comparison method using only the change in the water level before and after entering the bathtub. The proposed methods uses DTW distance of the waveform of entering water section and exiting water section and its average accuracy achieved 95%.

This paper presents a real-time scheduling framework for the robot operating system (ROS) called ROSCH. ROS is an open-source software platform and a meta-operating system designed for robots. ROS provides extensive development libraries and has been widely used for autonomous driving systems. However, ROS does not guarantee real-time performance; hence, a ROS-based autonomous driving car could cause a traffic accident. Therefore, ROSCH comprises three functionalities that do not exist in the ROS to guarantee real-time performance: (1) a synchronization system, (2) fixed-priority based directed acyclic graph (DAG) scheduling framework, and (3) fail-safe function. In particular, the synchronization system guarantees that the timestamp gaps between sensor measurements will be less than or equal to the calculated value. The fixed-priority based DAG scheduling framework guarantees that end-to-end latency is less than or equal to an estimated value. Operating both mechanisms simultaneously guarantees the final output topic frequency. The fail-safe functionality provides a danger avoidance action at a deadline miss. In addition, these functionalities are designed to be compatible with legacy software. No previous method that guarantees real-time operation provides a comparable range of features while maintaining the compatibility with the existing software. Experimental results showed that the time differences between sensor timestamps remained below a calculated value. Moreover, correctly scheduled tasks incur an end-to-end latency that is within an estimated value. In addition, the throughput of the end node is achieved. Finally, the fail-safe functionality overhead is 45 μsec on an average, which is 0.004% of the average execution time for one node.

The technology for detection of indoor location information has been attracting much attention in the recent past, since such information can be utilized for navigation and marketing research. The commonly used indoor location detection methods are Wi-Fi positioning, Bluetooth Low Energy (BLE) beacons and Pedestrian Dead-reckoning (PDR). PDR can provide location information relative to the original position by analyzing acceleration and angular velocity using sensors in smartphones without needing any positioning equipment in the environment. Meanwhile, to know the user's context in the indoor environment, PDR, RFID and surveillance cameras are investigated. Although PDR results can provide the atomic physical behavior such as the direction of walking (forward/backward or right/left), it is difficult to detect the user's shopping behaviors to determine if he/she is approaching, searching or just browsing in the shopping venues. Our research objective is to estimate the user's walking context appropriate for shopping venues with the same sensor information as used for conventional PDR. We propose a new method for walking context recognition that uses machine learning with a feature vector of accelerometers and gyroscopes. Based on the experiments conducted, we found that this model was able to recognize the intuitive walking context. We also established the utility of our proposal in the experiment at the actual shopping venue and verified that it can be applied to improve PDR positioning precision by 8%.

Graphics processing unit (GPU) programming environments have matured for general-purpose computing on GPUs. Significant challenges for GPUs include system software support for bounded response times and guaranteed throughput. In recent years, GPU technologies have been applied to real-time systems by extending the operating system modules to support real-time GPU resource management. Unfortunately, such a system extension makes it difficult to maintain the system with version updates because the OS kernel and device drivers must be modified at the source-code level, thereby preventing continuous research and development of GPU technologies for real-time systems. A loadable kernel module (LKM) framework, called Linux Real-Time eXtention with GPUs (Linux-RTXG), for managing real-time GPU resources with Linux without modifying the OS kernel and device drivers is proposed and evaluated experimentally. Linux-RTXG provides mechanisms for interrupt interception and independent synchronization to achieve real-time scheduling and resource reservation capabilities for GPU applications on top of existing device drivers and runtime libraries. Experimental results demonstrate that the overhead incurred by introducing the proposed Linux-RTXG is comparable to that of introducing existing kernel-dependent approaches. In addition, the results demonstrate that multiple GPU applications can be scheduled successfully by Linux-RTXG to meet their priority and quality-of-service requirements in real time.

Robot technologies, such as robot technology middleware (RTM) that is a component-oriented platform, are popular. However, RTM does not ensure stable real-time processing in common object request broker architecture. In this paper, a collaboration framework of RTM and TOPPERS embedded component system (TECS) is proposed to address this problem. TECS, a system that satisfies real-time processing requirements, is employed to enhance real-time processing in the proposed framework. To implement the collaboration of RTM and TECS, we have adopted remote procedure call and one-way communication. In addition, extending a generator enables the generation of robot technology components from TECS components. We have evaluated the processor cycle counts of the proposed framework in comparison with those of a conventional method. In addition, we evaluated the execution time of serial communication and a motor application using the proposed framework. The evaluation results show that the proposed framework is functionally employed in a hard real-time system. Furthermore, we evaluated the amount of code generated by the proposed framework. The evaluation results reveal that the code generated by the proposed framework is reusable and can enhance productivity.

Scripting languages are attractive for embedded system due to their high productivity. However, it is difficult to use scripting languages in a practical application because their performance and libraries for managing embedded devices are immature compared to the C programming language. This paper proposes a framework for effectively running an mruby script program on embedded systems based on the TOPPERS Embedded Component System (TECS). TECS generates glue code for invocation from mruby programs to legacy code in C language. It also supports configuration of mruby. Experimental results demonstrate the effectiveness of the proposed framework.

Graphics processing units (GPUs) embrace many-core compute devices where massively parallel compute threads are offloaded from CPUs. This heterogeneous nature of GPU computing raises non-trivial data transfer problems especially against latency-critical real-time systems. However even the basic characteristics of data transfers associated with GPU computing are not well studied in the literature. In this paper, we investigate and characterize currently-achievable data transfer methods of cutting-edge GPU technology. We implement these methods using open-source software to compare their performance and latency for real-world systems. Our experimental results show that the hardware-assisted direct memory access (DMA) and the I/O read-and-write access methods are usually the most effective, while on-chip micro controllers inside the GPU are useful in terms of reducing the data transfer latency for concurrent multiple data streams. We also disclose that CPU priorities can protect the performance of GPU data transfers.

We have analyzed logs that which web pages are viewed by users. Information recommendation is one of hot research areas for users activities support. Although many of recommendation systems are eager to match a user's preference, if the user does not want that at that moment, it would be just a noise no matter how much match the preference matches user's preference over all. It is important to understand what the user really wants each of moment timely. Therefore, in this paper, we make use of the following two characteristics for inference user's temporal wish. First is to adapt the degree of user's each interest with time range evolution. Second, web browsing logs related to an activity has been temporarily reinforced. The preliminary result of an algorithm is introduced.

Recently, RTM (Robot Technology Middleware) is attracting attention as a component oriented platform for robot development. However, RTM is unable to ensure real-time processing in CORBA because CORBA manages packets in a FIFO manager. In this paper, we propose a communication method from RTM to TECS in an effort to enhance realtime processing. TECS is a component system for embedded systems and suitable for real-time systems. We remove a part of real-time processing in RTM. Moreover, TECS is added to enhance real-time processing because TECS support a real-time processing requirements. In addition, it is possible to generate the components to communicate from RTM to TECS by using a plug-in. In the evaluation, cycle counts between RTM and TECS are compared with those between RTM and RTM. Furthermore, the amount of codes which are generated code with the purpose method and written code by developers are compared.

Human activity recognition through the wearable sensor will enable a next-generation human-oriented ubiquitous computing. However, most of research on human activity recognition so far is based on small number of subjects, and non-public data. To overcome the situation, we have gathered 4897 accelerometer data with 116 subjects and compose them as HASC2011corpus. In the field of pattern recognition, it is very important to evaluate and to improve the recognition methods by using the same dataset as a common ground. We make the HASC2011corpus into public for the research community to use it as a common ground of the Human Activity Recognition. We also show several facts and results of obtained from the corpus.