• Proceedings of the KIEE Conference
• /
• 2015.07a
• /
• pp.1351-1352
• /
• 2015

본 논문에서는 지능형 영상 감시 시스템에서 보행자를 검출하고 추적을 수행하기 위해 은닉층 활성함수에 가우시안 대신 FCM를 사용한 RBFNNs 패턴분류기와 객체 추적 알고리즘인 Mean Shift를 융합한 시뮬레이터를 개발한다. 시뮬레이터는 검출부과 추적부로 나누며, 검출부에서는 입력 영상으로부터 기울기의 방향성을 이용한 HOG(Histogram of Oriented Gradient) 특징을 구하고 빠른 처리속도를 위해 PCA 알고리즘을 통해 차원수를 축소하고 pRBFNNs 패턴분류기를 통해 보행자를 검출 한다. 다음 추적부에서 객체 추적 알고리즘인 Mean Shift를 이용하여 검출된 보행자 추적을 수행한다.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2016.10a
• /
• pp.143-144
• /
• 2016

본 논문은 보행자의 각 발에 부착된 2개의 IMU(Inertial Measurement Unit) 정보를 융합하여 위치 추적 성능을 향상시키는 보행자 추측 항법 알고리즘을 제안하였다. 센서내의 방향드리프트로 인해 IMU기반 보행자 위치추적은 시간이 지남에 따라 성능이 크게 저하된다. 제안하는 알고리즘은 방향 드리프트로 인해 각 발의 이동경로가 발산하는 점에 착안하여, 보폭이 일정 값을 초과할 시 이를 보정하고 사용자의 위치를 계산한다. 실험을 통해 제안하는 알고리즘이 방향 드리프트를 효과적으로 감소시키는 것을 확인하였다.

• The Journal of The Korea Institute of Intelligent Transport Systems
• /
• v.15 no.6
• /
• pp.71-79
• /
• 2016

Evaluation of the walkability of the urban pedestrian network requires construction of a pedestrian network model that reflects Space Syntax Theory. Space Syntax Theory deduces an integration value through which materials for evaluation of the pedestrian network's connectivity can be produced; and can aid in illustrating the ease of walkability through the model's calculation of pedestrian indices such as movability, comfort, and safety. But the representation of space syntax theory in the pedestrian network requires that turn delay be added by means of a network-type construction method. While tree-based Dial Algorithm proposed for the logit-based probability walkability distribution model may be effective for link-based pedestrian volume distribution, it requires further network expansion to reflect turn delays. In this research, Vine-based Dial Algorithm is executed in order to obtain a measure reflecting the integration value for Space Syntax Theory. The Vine-based Dial Algorithm of two adjacent links, which forms the minimum unit of the Vine network, has the advantage of encompassing turn delay, and thus eliminates the need for network expansion. Usage of the model to evaluation of complicated pedestrian spheres such as urban roads is left to further research. Especially the progression of the proposed method is deduced through case study.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.39C no.11
• /
• pp.1000-1008
• /
• 2014

Situation awareness for vehicles and pedestrians is very critical to ensure safety. While on-board sensors or systems can easily detect line-of-sight pedestrians, it is difficult to locate the positions of out-of-sight pedestrians especially with no GPS service. This paper proposes a method for accurate and on-time localization of indoor pedestrians by nearby vehicles. The proposed method is based on mobile ad-hoc networks among vehicles and pedestrians, without relying on infrastructures such as GPS, WiFi AP, and Bluetooth-based systems. Also, this paper develops a genetic algorithm to accurately and promptly locate pedestrians. Finally, simulation results are presented to quantitatively evaluate the proposed method compared to other studies.

• Proceedings of the KIEE Conference
• /
• 2015.07a
• /
• pp.1345-1346
• /
• 2015

본 논문에서는 보행자 및 배경 이미지로부터 HOG-PCA 특징을 추출하고 다항식 기반 RBFNNs(Radial Basis Function Neural Network) 패턴분류기과 최적화 알고리즘을 이용하여 보행자를 검출하는 시스템 설계를 제안한다. 입력 영상으로부터 보행자를 검출하기 위해 전처리 과정에서 HOG(Histogram of oriented gradient) 알고리즘을 통해 특징을 추출한다. 추출된 특징은 고차원이므로 패턴분류기 분류 시 많은 연산과 처리속도가 따른다. 이를 개선하고자 PCA (Principal Components Analysis)을 사용하여 저차원으로의 차원 축소한다. 본 논문에서 제안하는 분류기는 pRBFNNs 패턴분류기의 효율적인 학습을 위해 최적화 알고리즘인 PSO(Particle Swarm Optimization)을 사용하여 구조 및 파라미터를 최적화시켜 모델의 성능을 향상시킨다. 사용된 데이터로는 보행자 검출에 널리 사용되는 INRIA2005_person data set에서 보행자와 배경 영상을 각각 1200장을 학습 데이터, 검증 데이터로 구성하여 분류기를 설계하고 테스트 이미지를 설계된 최적의 분류기를 이용하여 보행자를 검출하고 검출률을 확인한다.

• Spatial Information Research
• /
• v.18 no.5
• /
• pp.133-142
• /
• 2010

Many pedestrian or fire evacuation models have been studied last decades for modeling evacuation behaviors or analysing building structures under emergency situations. However, currently developed models do not consider the differences of visibility of pedestrians by obstacles such as furniture, wall, etc. The visibility of pedestrians is considered one of the important factors that affect the evacuation behavior, leading to making simulation results more realistic. In order to incorporate pedestrian's visibility into evacuation simulation, we should be able to give different walking speeds according to differences of visibility. We improved the existing floor field model based on cellular automata in order to implement the visibility. Using the space syntax theory, we showed how we split the indoor spaces depending on the different visibilities created by different levels of structural depths. Then, we improved the algorithm such that pedestrians have different speeds instead of simultaneous movement to other cells. Also, in order for developing a real time simulation system integrated w ith indoor sensors later, we present a process to build a 3D simulator using a spatial DBMS. The proposed algorithm is tested using a campus building.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.21 no.1
• /
• pp.42-48
• /
• 2011

This paper presents the vibration reduction algorithm at the walking-will recognition sensors on the uneven terrain. Recently, concern about walking assistant aids is increasing according to the increase in population of elder and handicapped person. However, most of walking aids don't have any actuators for its movement. So, general walking aids have weakness for its movement to upward/download direction of slope. To overcome the weakness of the general walking aids, many researches for active type walking aids are being progressed. Especially, vibration analysis and impulse reduction are one of the important elements of the active-type walking aid during moving on the outdoor area because the ground has many kinds of obstacles such as speed dumps, puddles and so on. So, we analyze the influence from vibration by uneven terrain. And then, we propose the impulse reduction algorithm to overcome the vibration. All the processes are verified experimentally in an active-type walking aid.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2014.11a
• /
• pp.164-166
• /
• 2014

본 논문에서는 HOG 기반 보행자 검출 및 추적에서, HOG feature의 슬라이딩 윈도우의 수와 피라미드 층 수가 알고리즘의 수행속도와 직접적인 관계가 있다는 것을 확인한다. 그리고 이 결과를 바탕으로 윈도우의 수와 피라마드 층 수를 줄이는 방법을 제안하여 전체적인 보행자 검출 및 추적 속도를 증가시키고자 한다. 구체적으로, 제안하는 알고리즘은 검출 단계에서 색상의 선명도를 이용하여 관심 영역을 프레임 내에 지정함으로써 슬라이딩 윈도우의 수를 줄이고, 부가적으로 피라미드 층 수 또한 줄어들어서 보행자 검출 속도를 향상시킨다. 그리고 추적 단계에서는 보행자로 검출된 윈도우의 색상 정보를 이용하여 검출된 보행자를 빠르고 정확하게 추적하는 하는 방법을 제시한다.

• The Journal of the Korea institute of electronic communication sciences
• /
• v.10 no.9
• /
• pp.1039-1048
• /
• 2015

In this paper, we propose an efficient stair locomotion method for a quadruped robot with mechanism of insectile legs using genetic algorithm(GA). In the proposed method, we first define the factors and the reachable region for the stair locomotion. In addition, we set the gene and the fitness function for GA and generate the gait trajectory by searching the landing position of a quadruped robot, which has the minimun distance of movement and the optimal energy stability margin(ESM). Finally, we verify the effectiveness and superiority of the proposed stair locomotion method through the computer simulations.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.18 no.4
• /
• pp.566-571
• /
• 2008

In this paper, we propose a footstep planning method of biped robot based on the Particle Swarm Optimization(PSO). We define configuration and locomotion primitives for biped robots in the 2 dimensional workspace. A footstep planning method is designed using learning process of PSO that is initialized with a population of random objects and searches for optima by updating generations. The footstep planner searches for a feasible sequence of locomotion primitives between a starting point and a goal, and generates a path that avoids the obstacles. We design a path optimization algorithm that optimizes the footstep number and planning cost based on the path generated in the PSO learning process. The proposed planning method is verified by simulation examples in cluttered environments.