To address the OPC UA transformation requirements of conventional factory data collection systems, we propose an innovative method for the automated construction of OPC UA information models utilizing graph databases. This method not only streamlines the process but also minimizes the need for extensive manual labor and specialized expertise. Drawing from the resemblance between OPC UA information models and OWL ontologies, we have devised mapping rules that facilitate the translation of knowledge graph data into OPC UA format. Leveraging the existing Neo4j database within the factory as the primary information source, we've developed an OWL ontology construction module for exporting ontology files. In parallel, an information model construction module has been designed to convert the ontology file into an OPC UA XML file, complete with the OPC UA information model. This XML file adheres to the official OPC UA specification, ensured through the use of the UA-ModelCompiler tool. To validate the effectiveness and viability of this construction method, we conducted functional tests and evaluations using a publicly available database. The results of these tests confirm the feasibility of our approach, marking a significant advancement in OPC UA information model construction for factory data systems.
In response to the problems of high energy consumption, short communication distance, and unstable network topology of wireless sensor networks in complex forest environments, this paper proposes a monitoring system for complex forest environments. The proposed system utilizes LoRa technology to design a wireless sensor network that enables low power consumption and long-distance transmission. In addition, a self-organizing network protocol is employed to ensure stable and reliable data transmission in the forest area. Meanwhile, a cloud platform is developed to receive, analyze, and visualize forest monitoring data in real-time, as well as to provide early warning information. The field test results show that the system can effectively capture real-time changes in environmental factor parameters within the forest area. The technical features of the proposed system, such as low power consumption, self-organizing network, stable and reliable communication, and long transmission distance, make it suitable for forest fire monitoring in harsh environments.
Image inpainting is an important image processing application in repairing images with damaged or undesirable contents. While traditional approaches based on pixel or patch matching lack the ability to generate novel contents for the missing regions in an image, generative neural network has shown its potential in making up meaningful contents for the masked areas. On the other hand, to bridge the gaps between visible and missing regions, both traditional and deep-learning based approaches utilize self-similarity property of an image. In generative learning, this self-similarity property is explored using attention mechanisms. Since structural information may have multiple meanings depending on neighborhood context but a single attention mechanism can only provide one interpretation, applying transformer architecture with multiple attentions in image inpainting should further improve the final results by giving multiple interpretations to self-similarity information. The limitation of applying transformer is that complexity of an attention mechanism is proportional to the square of the number of pixels in an image. This complexity could grow exponentially and become infeasible for capacity-limited hardware if image resolution is high. In this paper, we propose to replace spatial-attention based transformer with one based on channel attention. Thereby, the complexity is reduced to the square of the number of channels which does not depend on spatial resolution and can be easily controlled as a network parameter. To remedy the loss of spatial information because of channel attention, a spatial gating mechanism is added after the channel-attention transformer. After this new lightweight transformer architecture generates the coarse result, an attention-based refinement network is used to generate the final result. Experimental results show that the new transformer architecture can produce inpainting results comparable to state-of-the-art approaches with far less network parameters.
The emergency disaster response headquarters (HQ) is responsible for incrementally collecting disaster information on the whole area in a region. Our previous work studied a routing problem for patrolling vehicles to monitor such information along all streets and bring it to HQ multiple times on the way individually. The optimal routes for vehicles can be found by a systematic search so as to minimize the average delay time in incremental information collection. Further, we have also studied an area-segmentation approach in which each vehicle only collects the information in one sub-area. However, in such individual collection approaches, multiple (duplicated) transits on the same street by vehicles could happen frequently in general because each vehicle must return to HQ every time it brings the monitored information by itself for incremental collection. Therefore, this study explores a collaborative collection scheme where multiple vehicles work together instead of each vehicle bringing its own monitored information to HQ individually. In the proposed scheme, a vehicle can drop off the monitored information at a Pickup Point (PP) on the way rather than HQ. Then, another vehicle passing the PP picks up the stored information and brings it to HQ. This scheme helps to reduce the time taken to collect information by multiple vehicles because it can increase the amount of information brought by a single return to HQ at the expense of the deployment cost of PP. We also provide a systematic search to find the best routing for the collaborative collection scheme. Our simulation results indicate that two vehicles collaborating on information collection with only a single PP can clearly reduce both the average delay time in incremental collection and the completion time to collect all information, compared with those working individually.
Finding parking spaces is a time-consuming and fuel-wasting task, especially in crowded areas. To address this issue, researchers have proposed an automatic parking detection system using deep learning-based cameras in outdoor parking lots with sufficient lighting. However, the development of parking detection systems for indoor areas with low lighting has not yet been extensively researched. This paper proposed an automatic vehicle parking availability system for an indoor environment using an object detection algorithm and clustering methods. The proposed method was tested with parking spots on the ground floor of the Faculty of Science building on the at Kasetsart University Sriracha campus, Thailand. Six models that combined three object detection algorithms with two clustering methods were tested for performance evaluation. The results showed that the model combining YOLOV4 and OPTICS achieved the highest detection accuracy for occupied parking slots of about 95 percent.
