• 전자공학회논문지
• /
• 제52권8호
• /
• pp.132-139
• /
• 2015

카메라를 이용하는 시각(visual) SLAM(Simultaneous Localization And Mapping)은 로봇의 위치 등을 파악하는데 널리 이용되고 있다. 일반적으로 시각 SLAM은 움직임이 없는 고정된 특징점을 대상으로 연속적인 시퀀스 상에서 카메라의 움직임을 추정한다. 따라서 이동하는 객체가 많이 존재하는 상황에서는 안정적인 결과를 기대하기 어렵다. 본 논문에서는 이동 객체가 많은 상황에서 스테레오 카메라를 이용한 SLAM을 안정화하는 방법을 제안한다. 먼저, 스테레오 카메라를 이용하여 깊이영상을 추출하고 옵티컬 플로우를 계산한다. 그리고 좌우 영상의 옵티컬 플로우를 이용하여 시차변화(disparity change)를 계산한다. 그리고 깊이 영상에서 사람과 같이 움직이는 객체에 대한 ROI(Region Of Interest)를 구한다. 실내 상황에서는 벽과 같은 정적인 평면들이 움직이는 영역으로 잘못 판단되는 경우가 자주 발생한다. 이런 문제점을 해결하기 위해 깊이 영상을 X-Z 평면으로 사영하고 허프(hough) 변환하여 장면을 구성하는 평면을 결정한다. 앞의 과정에서 판단된 이동 객체 중에서 벽과 같은 장면 요소를 제외한다. 제안된 방법을 통해 정적인 특징점이 요구되는 SLAM의 성능을 보다 안정화할 수 있음을 확인하였다.

• 전자공학회논문지SC
• /
• 제49권4호
• /
• pp.90-101
• /
• 2012

단일 카메라 기반의 SLAM(Simultaneous Localization and Mapping)을 성공적으로 수행하기 위해서는 표식 선택이 매우 중요하다. 특히, 미지의 환경에서는 표식에 대한 사정정보가 없기 때문에 표식을 자동 선택하는 기술이 필요하다. 본 논문에서는 표식을 자동 선택하기 위해 인간의 시각 집중 방식을 모델링한 시각 집중 시스템을 이용한다. 기존의 시각 집중 시스템에서 윤곽선(Edge)는 시각 집중을 위한 중요한 요소 중 하나이다. 하지만 복잡한 실내 환경에서 윤곽선의 응답을 사용할 경우 정규화 연산으로 인해 정보가 많은 복잡한 영역의 윤곽선에 대한 응답은 낮아지고 특징이 없는 평면이나 평면들 간의 경계에서 높은 값을 가지게 된다. 또한 네 방향에 대한 응답 값을 사용하기 때문에 특징의 차원수가 증가해서 연산량도 증가한다. 본 논문에서는 앞에서 언급한 문제점들을 해결하기 위해 모서리 특징의 사용을 제안한다. 모서리 특징을 사용함으로써 정보가 많은 복잡한 영역을 우선 집중시켜 데이터 연관(Data association)의 정확도도 높일 수 있다. 최종적으로는 코너특징을 사용한 시각 집중 시스템을 이용함으로써 기존 방식보다 SLAM 결과가 향상 된다는 것을 실험으로 보이도록 하겠다.

• ETRI Journal
• /
• 제43권4호
• /
• pp.603-616
• /
• 2021

Over the last few years, autonomous vehicles have progressed very rapidly. The odometry technique that estimates displacement from consecutive sensor inputs is an essential technique for autonomous driving. In this article, we propose a fast, robust, and accurate odometry technique. The proposed technique is light detection and ranging (LiDAR)-based direct odometry, which uses a spherical range image (SRI) that projects a three-dimensional point cloud onto a two-dimensional spherical image plane. Direct odometry is developed in a vision-based method, and a fast execution speed can be expected. However, applying LiDAR data is difficult because of the sparsity. To solve this problem, we propose an SRI generation method and mathematical analysis, two key point sampling methods using SRI to increase precision and robustness, and a fast optimization method. The proposed technique was tested with the KITTI dataset and real environments. Evaluation results yielded a translation error of 0.69%, a rotation error of 0.0031°/m in the KITTI training dataset, and an execution time of 17 ms. The results demonstrated high precision comparable with state-of-the-art and remarkably higher speed than conventional techniques.

• 한국측량학회지
• /
• 제37권2호
• /
• pp.45-53
• /
• 2019

RGB-D SLAM (Simultaneous Localization and Mapping) refers to the technology of using deep camera as a visual sensor for SLAM. In view of the disadvantages of high cost and indefinite scale in the construction of maps for laser sensors and traditional single and binocular cameras, a method for creating three-dimensional map of indoor environment with deep environment data combined with RGB-D SLAM scheme is studied. The method uses a mobile robot system equipped with a consumer-grade RGB-D sensor (Kinect) to acquire depth data, and then creates indoor three-dimensional point cloud maps in real time through key technologies such as positioning point generation, closed-loop detection, and map construction. The actual field experiment results show that the average error of the point cloud map created by the algorithm is 0.0045m, which ensures the stability of the construction using deep data and can accurately create real-time three-dimensional maps of indoor unknown environment.

• 제어로봇시스템학회논문지
• /
• 제19권9호
• /
• pp.813-821
• /
• 2013

In this paper, we present the development of P-SURO II hybrid AUV (Autonomous Underwater Vehicle) which can be operated in both of AUV and ROV (Remotely Operated Vehicle) modes. In its AUV mode, the vehicle is supposed to carry out some of underwater missions which are difficult to be achieved in ROV mode due to the tether cable. To accomplish its missions such as inspection and maintenance of complex underwater structures in AUV mode, the vehicle is required to have high level of autonomy including environmental recognition, obstacle avoidance, autonomous navigation, and so on. In addition to its systematic development issues, some of algorithmic issues are also discussed in this paper. Various experimental studies are also presented to demonstrate these developed autonomy algorithms.

• 한국군사과학기술학회지
• /
• 제18권6호
• /
• pp.711-720
• /
• 2015

Recently many studies of Radar systems mounted on ground vehicles for autonomous driving, SLAM (Simultaneous localization and mapping) and collision avoidance have been reported. Since several pixels per an object may be generated in a close-range radar application, a width of an object can be estimated automatically by various signal processing techniques. In this paper, we tried to attempt to develop an algorithm to estimate obstacle width using Radar images. The proposed method consists of 5 steps - 1) background clutter reduction, 2) local peak pixel detection, 3) region growing, 4) contour extraction and 5)width calculation. For the performance validation of our method, we performed the test width estimation using a real data of two cars acquired by commercial radar system - I200 manufactured by Navtech. As a result, we verified that the proposed method can estimate the widths of targets.

• 제어로봇시스템학회논문지
• /
• 제19권10호
• /
• pp.907-914
• /
• 2013

A FastSLAM is an algorithm for SLAM (Simultaneous Localization and Mapping) using a Rao-Blackwellized particle filter and its performance is known to degenerate over time due to the loss of particle diversity, mainly caused by the particle depletion problem in the resampling phase. In this paper, the GeSPIR (Geographically Stratified Particle Information-based Resampling) technique is proposed to solve the particle depletion problem. The proposed algorithm consists of the following four steps : the first step involves the grouping of particles divided into K regions, the second obtaining the normal weight of each region, the third specifying the protected areas, and the fourth resampling using regional equalization weight. Simulations show that the proposed algorithm obtains lower RMS errors in both robot and feature positions than the conventional FastSLAM algorithm.

• 제어로봇시스템학회논문지
• /
• 제22권12호
• /
• pp.1021-1026
• /
• 2016

The graph-based SLAM (Simultaneous Localization and Mapping) approach has been gaining much attention in SLAM research recently thanks to its ability to provide better maps and full trajectory estimations when compared to the filtering-based SLAM approach. Even though graph-based SLAM requires batch processing causing it to be computationally heavy, recent advancements in optimization and computing power enable it to run fast enough to be used in real-time. However, data association problems still require large amount of computation when building a pose graph. For example, to find loop closures it is necessary to consider the whole history of the robot trajectory and sensor data within the confident range. As a pose graph grows, the number of candidates to be searched also grows. It makes searching the loop closures a bottleneck when solving the SLAM problem. Our approach to alleviate this bottleneck is to sample a limited number of pose nodes in which loop closures are searched. We propose a heuristic for sampling pose nodes that are most advantageous to closing loops by providing a way of ranking pose nodes in order of usefulness for closing loops.

• 한국BIM학회 논문집
• /
• 제10권1호
• /
• pp.1-8
• /
• 2020

In managing building facilities, spatial information is the basic data for decision making. However, the method of acquiring spatial information is not easy. In many cases, the site and drawings are often different due to changes in facilities and time after construction. In this case, the site data should be scanned to obtain spatial information. The scan data actually contains spatial information, which is a great help in making space related decisions. However, to obtain scan data, an expensive LiDAR (Light Detection and Ranging) device must be purchased, and special software for processing data obtained from the device must be available.Recently, SLAM (Simultaneous localization and mapping), an advanced map generation technology, has been spreading in the field of robotics. Using SLAM, 3D spatial information can be obtained quickly in real time without a separate matching process. This study develops and tests whether SLAM technology can be used to obtain spatial information for facility management. This draws considerations for developing a SLAM device for real-time remote scanning for facility management. However, this study focuses on the system development method that acquires spatial information necessary for facility management through SLAM technology. To this end, we develop a prototype, analyze the pros and cons, and then suggest considerations for developing a SLAM system.

• 제어로봇시스템학회논문지
• /
• 제15권6호
• /
• pp.619-625
• /
• 2009

FastSLAM which uses the Rao-Blackwellized particle filter is one of the famous solutions to SLAM (Simultaneous Localization and Mapping) problem that estimates concurrently a robot's pose and surrounding environment. However, the particle depletion problem arises from the loss of the particle diversity in the resampling process of FastSLAM. Then, the performance of FastSLAM degenerates over the time. In this work, DIR (Density Information-based Resampling) technique is proposed to solve the particle depletion problem. First, the cluster is constructed based on the density of each particle, and the density of each cluster is computed. After that, the number of particles to be reserved in each cluster is determined using a linear method based on the distance between the highest density cluster and each cluster. Finally, the resampling process is performed by rejecting the particles which are not selected to be reserved in each cluster. The performance of the DIR proposed to solve the particle depletion problem in FastSLAM was verified in computer simulations, which significantly reduced both the RMS position error and the feature error.