• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.17 no.3
• /
• pp.861-880
• /
• 2023

Semantic segmentation of road scene is the key technology of autonomous driving, and the improvement of convolutional neural network architecture promotes the improvement of model segmentation performance. The existing convolutional neural network has the simplification of learning knowledge and the complexity of the model. To address this issue, we proposed a road scene semantic segmentation algorithm based on multi-task collaborative learning. Firstly, a depthwise separable convolution atrous spatial pyramid pooling is proposed to reduce model complexity. Secondly, a collaborative learning framework is proposed involved with saliency detection, and the joint loss function is defined using homoscedastic uncertainty to meet the new learning model. Experiments are conducted on the road and nature scenes datasets. The proposed method achieves 70.94% and 64.90% mIoU on Cityscapes and PASCAL VOC 2012 datasets, respectively. Qualitatively, Compared to methods with excellent performance, the method proposed in this paper has significant advantages in the segmentation of fine targets and boundaries.

• Current Research on Agriculture and Life Sciences
• /
• v.19
• /
• pp.45-54
• /
• 2001

This study was carried out to provide fundamental data for prospective forest-road project and forest-road network arrangement through appraising existing forest-road network with density, extension distance, maximum yarding distance and yarding area, position of forest-road line considered above foundation of two theories, one is "theory of optimal forest-road density" which has expense for yarding cost and constructing forest-road minimized, the other is "theory of optimal forest-road arrangement" which has investment effect maximized. The results are as follows. 1. In density and extension distance of the forest-road by site, it was showed up that density of existing forest-road is lower than that of calculated forest-road. So, it is thought that some additional forest-roads have to be constructed. 2. In the arrangement of the forest-road network by site, it was showed up that the arrangement of calculated forest-road is higher than that of existing forest-road arrangement for the forestry and yarding function. So, it is thought that the arrangement of forest-road network have to be considered to maximize the investment effect. 3. In "mean maximum distance for yarding" and "mean area which yarding can be done" by horizontal and inclined distance, the existing forest-road networks were different from those of calculated forest-road network. So, calculated forest-road network making investment effect maximize is more effective than existing forest-road network. Hence, in prospective forest-road project, it is needed that forest-road network having "area which yarding can be done" maximized through considering function for yarding have to be constructed.

• 한국ITS학회:학술대회논문집
• /
• 2003.11a
• /
• pp.132-135
• /
• 2003

• Proceedings of the Korea Society for Simulation Conference
• /
• 1997.04a
• /
• pp.81-85
• /
• 1997

In a virtual environment, agents must demonstrate some degree of realism and interactivity. This paper discusses the algorithm that enables agents to navigate a virtual road network realistically and interactively. The road description files written in this language provide the information of road environments to the navigating agents and the scene visualizer. We call this navigating agent in the road an ambient car. The ambient cars must follow the traffic rules as human does. To do this, the ambient car should continuously check its circumstances, such as, the traffic lights, lanes, road signs, and other ambient cars. Because of the huge scale of road network and the large number of ambient cars, the algorithm considers only the area where the participant is currently located. By this locality, the performance of the whole system does not fluctuate much in different situations. The behavior of ambient cars according to the predefined rules may appear monotonous. We added probability distribution functions to introduce some randomness. We implemented the above idea on silicon Graphics Indigo 2 workstation. The ambient car exhibited its awareness of lanes, traffic lights, and other cars. The participants could hardly distinguish between a human-controlled car and computer-controlled ambient car generated by the algorithm.

• 한국지구물리탐사학회:학술대회논문집
• /
• 2003.11a
• /
• pp.252-254
• /
• 2003

In this study, it was purposed to develop the new method for the prediction of pollutant concentration in road tunnels. The new method was the use of artificial neural network with the back-propagation algorithm which can model the non-linear system of tunnel environment. This network system was separated into two parts as the visibility and the CO concentration. For this study, data was collected from two highway road tunnels on Yeongdong Expressway. The tunnels have two lanes with one-way direction and adopt the longitudinal ventilation system. The actually measured data from the tunnels was used to develop the neural network system for the prediction of pollutant concentration. The output results from the newly developed neural network system were analysed and compared with the calculated values by PIARC method. Results showed that the prediction accuracy by the neural network system was approximately five times better than the one by PIARC method. ill addition, the system predicted much more accurately at the situation where the drivers have to be stayed for a while in tunnels caused by the low velocity of vehicles.

• Spatial Information Research
• /
• v.20 no.5
• /
• pp.57-66
• /
• 2012

Recent development in wireless communication technology like GPS as well as mobile equipments like PDA and cellular phone makes location-based services (LBSs) popular. However, because users request a query to LBS servers by using their exact locations while moving on the road network, users' privacy may not be protected in the LBSs. Therefore, a mechanism for users' privacy protection is required for the safe and comfortable use of LBSs by mobile users. For this, we, in this paper, propose a road network distance based cloaking scheme supporting user privacy protection in location-based services. The proposed scheme creates a cloaking area by considering road network distance, in order to support the efficient and safe LBSs on the road network. Finally, we show from our performance analysis that our cloaking scheme outperforms the existing cloaking scheme in terms of cloaking area and service time.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.8 no.3
• /
• pp.593-600
• /
• 2007

In this paper, we composed several road network and evaluated the network for the performance of the network with protocols. The protocols we applied were the MANET routing protocols such as AODV(Ad-hoc On-Demand Distance Vector) and DSR(Dynamic Source Routing) protocols. Generally, the AODV performs better than the DSR. However, in my ad-hoc vehicular network, the performance of the DSR is the better than the AODV when there are more vehicles in the road environment than there are the less vehicles. For the simulation, we composed 4-lane road with vehicles and simulated in the OPNET.

• Journal of Korean Society of Forest Science
• /
• v.105 no.1
• /
• pp.115-121
• /
• 2016

This study was conducted to provide the basic policy information for establishing efficient forest-road networks. Synthetic forest-road networks that consist of main and spur roads and forest-road networks with only main road (hereafter called "main-road network") were planned for the five forest-road experimental districts of Korea Forest Service in this study. Road density of the synthetic forest-road networks was calculated and compared with the road density of the main-road networks. The results showed that the optimum road density of the synthetic forest-road networks was 10.1~15.9 m/ha, and the road density of the main-road networks was 8.4~12.4 m/ha. The construction cost of the synthetic forest-road networks was estimated about 1~8% lower than the main-road networks, while the road density was 20~30% greater than the main-road networks. As timber volume and hauling cost increased, the optimum road density of the synthetic forest-road networks increased, within which the road density of highstandard main road rapidly increased. On the other hand, the spur road density increased with slope gradient.

• The Journal of The Korea Institute of Intelligent Transport Systems
• /
• v.19 no.2
• /
• pp.18-35
• /
• 2020

Many human, roadway, and vehicle factors affect vehicle-pedestrian crashes. Especially, the roadway factors are easily defined and suitable for suggesting countermeasures. The characteristics of the road network are one of the roadway factors. The road network significantly influences behaviors and conflicts of drivers and pedestrians. A metropolitan city such as Seoul contains various types of road networks, and crash prevention strategy considering characteristics of the road network is required. In this study, we analyze the effects of road networks on vehicle-pedestrian crashes. In the study, high order road ratio, intersection ratio, high-low intersection ratio are considered as road network variables. Using Geographically Weighted Poisson Regression, crash frequencies in Dongs of Seoul are analyzed based on the road network variable as well as socioeconomic variables. As a result, Dongs are grouped by coefficient signs, and each group is suggested about improvement directions considering conflict situations.

• Journal of Korean Society of Transportation
• /
• v.17 no.3
• /
• pp.73-85
• /
• 1999

In metropolitan area, intersections having turn Prohibitions are often found and it is important to consider them in path planning. This Paper presents a new path planning method road networks having turn prohibitions. A new road network model is proposed in which an alter-native route for each turn-Prohibition is constructed in off-line using U-turn or P-turn and then Put into road network database. The proposed network model is efficient since it requires no virtual nodes that are usually used in the conventional road networks to represent turns at intersections. In order to find a shortest path with the proposed network, a new shortest path algorithm is proposed. A knot for the turn-Prohibited node is newly defined and is used in comparing the cost of the alternative route with that of the other path.