• Journal of Korea Society of Industrial Information Systems
• /
• v.27 no.4
• /
• pp.19-27
• /
• 2022

In general, the performance of ML(Machine Learning) application is determined by various factors such as the type of ML model, the size of model (number of parameters), hyperparameters setting during the training, and training data. In particular, the recognition accuracy of ML may be deteriorated or experienced overfitting problem if the amount of dada used for training is insufficient. Existing studies focusing on image recognition have widely used open datasets for training and evaluating the proposed ML models. However, for specific applications where the sensor used, the target of recognition, and the recognition situation are different, it is necessary to build the dataset manually. In this case, the performance of ML largely depends on the quantity and quality of the data. In this paper, training data used for motion recognition application is augmented using the kernel density estimation algorithm which is a type of non-parametric estimation method. We then compare and analyze the recognition accuracy of a ML application by varying the number of original data, kernel types and augmentation rate used for data augmentation. Finally experimental results show that the recognition accuracy is improved by up to 14.31% when using the narrow bandwidth Tophat kernel.

• The Journal of The Korea Institute of Intelligent Transport Systems
• /
• v.22 no.4
• /
• pp.194-210
• /
• 2023

LiDAR plays a key role in autonomous vehicles, where it is used to detect the environment in place of the driver's eyes, and its role is expanding. In recent years, there has been a growing need to test the performance of LiDARs installed in autonomous vehicles. Many LiDAR performance tests have been conducted in simulated and indoor(lab) environments, but the number of tests in outdoor(field) and real-world road environments has been minimal. In this study, we compared LiDAR performance under the same conditions lab and field to determine the relationship between lab and field tests and to establish the characteristics and roles of each test environment. The experimental results showed that LiDAR detection performance varies depending on the lighting environment (direct sunlight, led) and the detected object. In particular, the effect of decreasing intensity due to increasing distance and rainfall is greater outdoors, suggesting that both lab and field experiments are necessary when testing LiDAR detection performance on objects. The results of this study are expected to be useful for organizations conducting research on the use of LiDAR sensors and facilities for LiDAR sensors.

• Journal of the Korean Institute of Gas
• /
• v.27 no.1
• /
• pp.57-62
• /
• 2023

The Gas Regulator EVENT is a signal sent by the Remote Terminal Unit(RTU) installed in each local gas regulator (hereinafter referred to as "regulator"), and is an abnormal alarm that can be intuitively checked in our client server. This is an important data that enables immediate dispatch order and initial action in the situation room when a regolator abnormality occurs, and can analyze the cause of the regulator abnormality. Looking at the trend of EVENT data for regulator over the past three years, there is a clear trend of increasing unchecked EVENT data. The increase in non-checking event (actual abnormality or noise event) may mean that firstly, mechanical or pressure abnormality occurs in the actual regulator, and secondly, there is no abnormality in the actual regulator, but communication error occurred in the RTU, reset. EVENT Data may be formed as if an abnormality occurred in the static voltage due to an error, sensor error, power failure (instantaneous power failure), etc. Among them, this study analyzed the recently generated unchecked EVENT data to identify critical noise events among RTU errors, which are noise events, and reviewed ways to increase the reliability of Regulator EVENTs by reducing them.

• Smart Media Journal
• /
• v.12 no.1
• /
• pp.77-91
• /
• 2023

Interaction methods and technologies for mutual exploration based on user behavior are evolving variously. Especially, in recent years, with the development of a wide range of sensors, they have developed from contact to non-contact methods. However, developers' senseless definitions of the interaction methods have made the exploration process quite complicated, which rather creates the hassle of users needing to learn the interaction guide defined by the developers before experiencing the exhibition contents. In this context, in order to make visitors smoothly communicate with exhibition contents, a preliminary study on easy interaction for users of various ages is needed, and in particular, research on improving the usability of user interaction is also essential when developing non-contact exhibition contents. So, in this study, a method to reduce the confusion between developers and users was sought by researching non-contact interaction that could be universally interacted with in the field of exhibition contents and proposing behavior designs. First, based on the narrative structure of cultural resources, existing studies were reviewed and the points of interactions as cultural contents were derived. Then the most efficient search process was selected among non-contact behaviors based on hand gestures that allow users to naturally guess and learn interaction methods. Furthermore, on the basis of the meaning of non-linear narrative-based interaction and the analysis results of spatial behavior elements, affordance behavior with high learning effect and efficiency was derived. Through this research process, an action that helps users to understand non-contact interaction naturally in the process of exploring exhibition-type cultural contents and to utilize non-contact interaction in the process of immersion in exhibition contents is proposed as a final model.

• The Journal of Society for e-Business Studies
• /
• v.27 no.2
• /
• pp.205-218
• /
• 2022

Recently, digital transformation in manufacturing has been accelerating. It results in that the data collection technologies from the shop-floor is becoming important. These approaches focus primarily on obtaining specific manufacturing data using various sensors and communication technologies. In order to expand the channel of field data collection, this study proposes a method to automatically collect manufacturing data based on vision-based artificial intelligence. This is to analyze real-time image information with the object detection and tracking technologies and to obtain manufacturing data. The research team collects object motion information for each frame by applying YOLO (You Only Look Once) and DeepSORT as object detection and tracking algorithms. Thereafter, the motion information is converted into two pieces of manufacturing data (production performance and time) through post-processing. A dynamically moving factory model is created to obtain training data for deep learning. In addition, operating scenarios are proposed to reproduce the shop-floor situation in the real world. The operating scenario assumes a flow-shop consisting of six facilities. As a result of collecting manufacturing data according to the operating scenarios, the accuracy was 96.3%.

• Korean Journal of Remote Sensing
• /
• v.39 no.6_1
• /
• pp.1451-1466
• /
• 2023

In the context of maritime emergencies, the utilization of drones has rapidly increased, with a particular focus on their application in search and rescue operations. Deep learning models utilizing drone images for the rapid detection of distressed vessels and other maritime drift objects are gaining attention. However, effective training of such models necessitates a substantial amount of diverse training data that considers various weather conditions and vessel states. The lack of such data can lead to a degradation in the performance of trained models. This study aims to enhance the performance of deep learning models for distress ship detection by developing a maritime environment simulator to augment the dataset. The simulator allows for the configuration of various weather conditions, vessel states such as sinking or capsizing, and specifications and characteristics of drones and sensors. Training the deep learning model with the dataset generated through simulation resulted in improved detection performance, including accuracy and recall, when compared to models trained solely on actual drone image datasets. In particular, the accuracy of distress ship detection in adverse weather conditions, such as rain or fog, increased by approximately 2-5%, with a significant reduction in the rate of undetected instances. These results demonstrate the practical and effective contribution of the developed simulator in simulating diverse scenarios for model training. Furthermore, the distress ship detection deep learning model based on this approach is expected to be efficiently applied in maritime search and rescue operations.

• The Journal of The Korea Institute of Intelligent Transport Systems
• /
• v.23 no.2
• /
• pp.157-172
• /
• 2024

Autonomous driving technology, when commercialized, has the potential to improve the safety, mobility, and environmental performance of transportation networks. However, safe autonomous driving may be hindered by poor sensor performance and limitations in long-distance detection. Therefore, cooperative autonomous driving that can supplement information collected from surrounding vehicles and infrastructure is essential. In addition, since HDVs, AVs, and CAVs have different ranges of perceivable information and different response protocols, countermeasures are needed for mixed traffic that occur during the transition period of autonomous driving technology. There is a lack of research on traffic flow optimization that considers the penetration rate of autonomous vehicles and the different characteristics of each road segment. The objective of this study is to develop weights based on safety, operational, and environmental factors for each infrastructure control use case and autonomous vehicle MPR. To develop an integrated evaluation index, infra-guidance AHP and hybrid AHP weights were combined. Based on the results of this study, it can be used to give right of way to each vehicle to optimize mixed traffic.

• Journal of the Korean Association of Geographic Information Studies
• /
• v.27 no.1
• /
• pp.29-40
• /
• 2024

Recently, the technology of autonomous driving, one of the core of the fourth industrial revolution, is developing, but sensor-based autonomous driving is showing limitations, such as accidents in unexpected situations, To compensate for this, HD-map is being used as a core infrastructure for autonomous driving, and interest in the public and private sectors is increasing, and various studies and technology developments are being conducted to secure the latest and accuracy of HD-map. Currently, NGII will be newly built in urban areas and major roads across the country, including the metropolitan area, where self-driving cars are expected to run, and is working to minimize data error rates through quality verification. Therefore, this study analyzes the spatial relationship of reference objects in the attribute structuring process for rapid and accurate renewal and production of HD-map under construction by NGII, By applying the attribute input automation methodology of the reference object in which spatial relations are established using the library of open source-based PyQGIS, target sites were selected for each road type, such as high-speed national highways, general national highways, and C-ITS demonstration sections. Using the attribute automation tool developed in this study, it took about 2 to 5 minutes for each target location to automatically input the attributes of the spatial relationship reference object, As a result of automation of attribute input for reference objects, attribute input accuracy of 86.4% for high-speed national highways, 79.7% for general national highways, 82.4% for C-ITS, and 82.8% on average were secured.

• Journal of Industrial Convergence
• /
• v.22 no.7
• /
• pp.15-21
• /
• 2024

A hiking stick is generally one of the walking aids that allow hikers to walk while relying on their own bodies when walking. A rechargeable battery must be built into the hiking stick, which is an auxiliary device, in order to perform various functions. A separate power supply is required to charge the rechargeable battery. This study is about a self-generated power supply and develops a power generation device using a screw with higher power generation efficiency than the existing method. It is differentiated from the method suggested in this study by comparing and analyzing it with the existing power generation method, and identifying problems therewith. The screw-type power generation device generates power when the climbing stick comes into contact with the ground and when it is separated from the ground. The built-in power generation device does not require a separate power supply, and it can be used by attaching the role of a mobile phone auxiliary battery and a lighting lamp, and it has the effect of being able to find it through location tracking by embedding a GPS sensor, etc., and using lighting to keep the user safe in emergency situations such as distress. The existing generator with built-in mountain climbing stick is difficult to charge due to very weak current and low practicality, but the generator developed in this research could achieve high efficiency to obtain a sufficient current, so it is possible to charge a battery and practicality.

• Journal of Intelligence and Information Systems
• /
• v.17 no.2
• /
• pp.119-128
• /
• 2011

The Sensor technology and portable device capability able to collect recent user information and the information about the surrounding environment haven been highly developed. A user can be made use of various contents and the option is also extending with this technology development. In particular, the initial portable device had simply a call function, but now that has evolved into 'the 4th screen' which including movie, television, PC ability. also, in the past, a portable device to provided only the services of a SMS, in recent years, it provided to interactive video service, and it include technology which providing various contents. Also, it is rising as media which leading the consumption of contents, because it can be used anytime, anywhere. However, the contents available for the nature of user's handheld devices are limited. because it is very difficult for making the contents separately according to various device specification. To find a solution to this problem, the study on one contents from several device has been progressing. The contents conversion technology making use of the profile of device out of this study comes to the force and profile study has been progressing for this. Furthermore, Demand for a user is also increased and the study on the technology collecting, analyzing demands has been making active progress. And what is more, Grasping user's demands by making use of this technology and the study on the technology analyzing, providing contents has been making active progress as well. First of all, there is a method making good use of ZigBee, Bluetooth technology about the sensor for gathering user's information. ZigBee uses low-power digital radio for wireless headphone, wireless communication network, and being utilized for smart energy, automatic home system, wireless communication application and wireless sensor application. Bluetooth, as industry standards of PAN(Personal Area Networks), is being made of use of low power wireless device for the technology supporting data transmission such as drawing file, video file among Bluetooth device. With analyzing the collected information making use of this technology, it utilizes personalized service based on network knowledge developed by ETRI to service contents tailor-made for a user. Now that personalized service builds up network knowledge about user's various environments, the technology provides context friendly service constructed dynamically on the basis of this. The contents to service dynamically like this offer the contents that it converses with utilizing device profile to working well. Therefore, this paper suggests the system as follow. It collects the information, for example of user's sensitivity, context and location by using sensor technology, and generates the profile as a means of collected information as sensor. It collects the user's propensity to the information by user's input and event and generates profile in the same way besides the gathered information by sensor. Device transmits a generated profile and the profile about a device specification to proxy server. And proxy server transmits a profile to each profile management server. It analyzes profile in proxy server so that it selects the contents user demand and requests in contents server. Contents server receives a profile of user portable device from device profile server and converses the contents by using this. Original source code of contents convert into XML code using the device profile and XML code convert into source code available in user portable device. Thus, contents conversion process is terminated and user friendly system is completed as the user transmits optimal contents for user portable device.