The mobile robot which accomplishes a work in explored region does not know location information of surroundings. Traditionally, simultaneous localization and mapping() algorithms solve the localization and mapping problem in explored regions. Among the several algorithms, the EKF (Extended Kalman Filter) based is the scheme most widely used. The EKF is the optimal sensor fusion method which has been used for a long time. The odometric error caused by an encoder can be compensated by an EKF, which fuses different types of sensor data with weights proportional to the uncertainty of each sensor. In many cases the EKF based requires artificially installed features, which causes difficulty in actual implementation. Moreover, the computational complexity involved in an EKF increases as the number of features increases. And is a weak point of long operation time. Therefore, this paper presents a symmetrical model based algorithm (called M-).M- for experience based prediction becomes the help which shortens an operation time. Experience leads and composes hypotheses map. Will use hypotheses map where is composed and there is a possibility which will raise efficiency. When the area where is repeated exists plentifully the time is shortened.

抄録全体を表示

Recently,

(Simultaneous Localization and Mapping) becomes a hot research topic in computer vision due to high demand to reconstruct surrounding environment with camera mounted on drone, etc. In system, Direct system has reported that it can achieve high 3D reconstruction environment since it utilizes the information of all pixels in images. In addition, to create accurate 3D reconstruction environment, system requires high camera tracking performance. Fortunately, to achieve this, it is well-known that a camera with wider field of view can help the performance improvement. From this fact, in this paper, we propose a new Direct system with 360-degree camera, which has two fish-eye lens and can capture front and back view scene. The characteristic points of our proposed method is that 1) we utilize LSD- system, which is a Direct one, since it can achieve faster 3D reconstruction compared with other systems, 2) we integrate the information of two camera coordinates to achieve high camera tracking accuracy, 3) we cover the frame-in/frame-out information by seamlessly taking over the camera tracking information from a fish-eye camera system to another one. From the experimental results with CG simulation, we achieved higher camera tracking accuracy with 360-degree camera compared to that with a fish-eye camera.

抄録全体を表示

森林内での位置情報の取得は森林管理から素材生産まで幅広く利用可能な重要な技術である。しかし，森林内ではGNSS を利用してリアルタイムで正確な位置情報を取得することは困難である。特に近年注目されている林業機械の自動走行について考えると，リアルタイムでの位置情報の取得は必要不可欠である。本研究では林内の自己位置推定技術として測域センサおよびRGBD カメラによる

に着目し，精度の検討を行った。Robot Operating System 上で実装がある として，測域センサを利用したLOAM，hdl graph ，RGBD カメラを用いたVisual としてORB-2，RTAB-Map について検討を行った。それぞれ，水平方向のRMSE がLOAM では0.526 m，RTAB-Map では0.619 m という結果が得られ，他センサと統合することでGNSS を十分に使用できない林内環境においてLRF およびRGBD カメラを用いた は位置推定手法として十分に適用できる可能性があることが示唆された。

抄録全体を表示

Achieving practical and full-scale use of drones will require a transition from a first-person view (FPV) flight based on visual radio control to an autonomous wide-area, long-distance flight. However, technology that enables drones to fly autonomously over a wide area and long distances while feeding back positioning in ever-changing real-world environments is yet to be established. We aimed to develop a

system that combines ORB- and dense point clouds in the environment. The result of an evaluation experiment of the developed system using a simulator indicated that the estimated self-position was corrected using matching with the dense point cloud, and a system combining ORB- and the dense point cloud was developed. It was confirmed that a system combining ORB- and some dense points in the environment could be developed. We have achieved good results in basic operation trials, demonstrating the potential of both systems for practical use.

抄録全体を表示

Although image change detection (ICD) methods provide good detection accuracy for many scenarios, most existing methods rely on place-specific background modeling. The time/space cost for such place-specific models is prohibitive for large-scale scenarios, such as long-term robotic visual simultaneous localization and mapping (

). Therefore, we propose a novel ICD framework that is specifically customized for long-term . This study is inspired by the multi-map-based framework, where multiple maps can perform mutual diagnosis and hence do not require any explicit background modeling/model. We extend this multi-map-based diagnosis approach to a more generic single-map-based object-level diagnosis framework (i.e., ICD), where the self-localization module of , which is the change object indicator, can be used in its original form. Furthermore, we consider map diagnosis on a state-of-the-art deep convolutional neural network (DCN)-based system (instead of on conventional bag-of-words or landmark-based systems), in which the blackbox nature of the DCN complicates the diagnosis problem. Additionally, we consider a three-dimensional point cloud (PC)-based (instead of typical monocular color image-based) and adopt a state-of-the-art scan context PC descriptor for map diagnosis for the first time.

抄録全体を表示

Measles virus (MV) is a member of the Morbillivirus genus in the Paramyxoviridae family. Human signaling lymphocyte activation molecule (

) acts as a cellular receptor for MV. is expressed on immature thymocytes, activated lymphocytes, macrophages and mature dendritic cells. This distribution of is in accord with lymphotropism and immunosuppressive nature of MV. Canine distemper and rinderpest viruses, other members of the Morbillivirus genus, also use as receptors. Laboratory-adapted MV strains often use ubiquitously expressed CD46 as an alternative receptor through the amino acid change(s) in the receptor-binding hemagglutinin. Furthermore, MV can infect various cultured cells, albeit with very low efficiency, via - and CD46-independent pathway, which may also account for MV infection of – cells in vivo.

抄録全体を表示

Signaling lymphocytic activation molecule (; CDw150) is a 70 kDa glycoprotein. Signaling lymphocytic activation molecule is constitutively expressed on memory T cells, CD56⁺ T cells, a subset of T cell receptor γδ⁺ cells, immature thymocytes and, at low levels, on a proportion of peripheral blood B cells. Signaling lymphocytic activation molecule is rapidly upregulated on all T and B cells after activation. Engagement of by F(ab′)₂ fragments of an anti- monoclonal antibody (mAb A12) enhances antigen-specific T cell proliferation. In addition, mAb A12 was directly mitogenic for T cell clones and activated T cells. T cell proliferation induced by mAb A12 is independent of interleukin (IL)-2, IL-4, IL-12 and IL-15, but is cyclosporin A sensitive. Ligation of during antigen-specific T cell proliferation resulted in upregulation of interferon (IFN)-γ production, even by allergen-specific T helper cell (Th) 2 clones, whereas the levels of IL-4 and IL-5 production were only marginally affected. The mAb A12 was unable to induce IL-4 and IL-5 production by Th1 clones. Costimulation of skin-derived Der P 1-specific Th2 cells from patients with atopic dermatitis via resulted in the generation of a population of IFN-γ-producing cells, thereby reverting their phenotype to a Th0 pattern. Signaling lymphocytic activation molecule is a high-affinity self ligand mediating homophilic cell interaction. In addition, soluble enhances both T and B cell proliferation. Collectively, these data indicate that molecules act both as receptors and ligands that are not only involved in T cell expansion but also drive the expanding T cells during immune responses into the Th0/Th1 pathway. This suggests that signaling through plays a role in directing Th0/Th1 development.

抄録全体を表示

The normal system can only build the environment map without any semantic information, which is insufficient in some specific application. This paper studies about fusing object recognition into system, and aims at building a semantic system.

抄録全体を表示

This paper deals with large scale map building by using Multi-Robot in D- framework. D- is a decoupled solution to the problem. Specifically, it is demonstrated that each robot estimates a local map using EKF, and these local maps are merged into a global map. In this paper, we propose a new RLS-based algorithm for map merging in this D- framework. First, we transform local maps into relative information which is considered as measurements for the global map. Then, we update the state estimate by RLS considering the weighting of measurements, which is determined by error propagation from the EKF . The convergence of the error covariance matrix in this algorithm can be proven. In experimental results, we confirm the validity of the proposed algorithm and correctness of derived theorems for the convergence.

抄録全体を表示

A hopping rover is a robot that can move in low gravity planets by the characteristic motion called the hopping motion. For its autonomous explorations, the so-called

(Simultaneous Localization and Mapping) is a basic function. is the combination of estimating the position of a robot and creating a map of an unknown environment. Most conventional methods of are based on odometry to estimate the position of the robot. However, in the case of the hopping rover, the error of odometry becomes considerably large because its hopping motion involves unpredictable bounce on the rough ground on an unexplored planet. Motivated by the above discussion, this paper addresses a problem of finding an optimal movement of the hopping rover for the estimation performance of the . For the problem, we first set the model of the system for the hopping rover. The problem is formulated as minimizing the expectation of the estimation error at a pre-specified time with respect to the sequence of control inputs. We show that the optimal input sequence tends to force the final position to be not at the landmark but in front of the landmark, and furthermore, the optimal input sequence is constant on the time interval for optimization.

抄録全体を表示

森林内での車両位置情報は運搬の自動化などに利用可能な技術だがGNSSを使用する場合，森林内ではリアルタイムに正確な位置情報を取得することが困難な場所があり，車両位置情報の取得をGNSS以外でする必要がある。本研究では弊社が独自に開発しているLiDAR

を林業機械向けに適用し，その精度の検討を行った。LiDAR を林業機械向けにIMU，ホイールエンコーダ情報との統合とLiDAR のマッチング処理を粗い精度と細かい精度の二段階にすることでICTフォワーダの自己位置推定時に水平方向で最大誤差0.064 m，平均誤差0.020 m，を用いた追従走行においても水平方向で最大誤差0.34 m，平均誤差0.17 mという結果が得られ，GNSSを用いることが困難な森林環境における自己位置推定手法として適用できることを確認した。

抄録全体を表示

This paper proposes a self-taught classifier of gateways for hybrid . Gateways are detected and recognized by the self-taught classifier, which is a SVM classifier and self-taught in that its training samples are produced and labeled without user's intervention. Since the detection of gateways at the topological boundaries of an acquired metric map reduces computational complexity in partitioning the metric map into sub-maps as compared with previous hybrid approaches using spectral clustering methods, from O(2ⁿ) to O(n), where n is the number of sub-maps. This makes possible real time hybrid even for large-scale metric maps. We have confirmed that the self-taught classifier provides satisfactory consistency and computationally efficiency in hybrid through different experiments.

抄録全体を表示

とは自己位置を計測しながら障害物マップを作ることであり,ロボットや自律走行分野に広く応用されてきた.従来ではジャイロセンサーや車速センサーを等の手法が主流であるが,タイヤの空回りやスリップに対応できないため,近年画像処理に基づいたVisual の手法が注目を集めている.本研究では,複数台のカメラを接続した全方位カメラから取得した画像を用いたVisual を提案する.この手法は単眼カメラより範囲が広く,曲面ミラーを用いた全方位カメラより使える部分が多い画像を用いることでより正確な自己位置の計測を可能にしている.

抄録全体を表示

Three-dimensional (3D) surface reconstruction is used to solve the problem of the narrow field of view in laparoscopy. It can provide surgeons or computer-assisted surgery systems with real-time complete internal abdominal anatomy. However, rapid changes in image depth, less texture, and specular reflection pose a challenge for the reconstruction. It is difficult to stably complete the reconstruction process using feature-based simultaneous localization and mapping (

) method. This paper proposes a robust laparoscopic 3D surface reconstruction method using , which can automatically select appropriate parameters for stereo matching and robustly find matching point pairs for laparoscope motion estimation. The changing trend of disparity maps is used to predict stereo matching parameters to improve the quality of the disparity map. Feature patch extraction and tracking are selected to replace feature point extraction and matching in motion estimation, which reduces its failure and interruption in feature-based . The proposed feature patch matching method is suitable for parallel computing, which can improve its computing speed. Evaluation results on public in vivo and ex vivo porcine abdominal video data show the efficiency and robustness of our 3D surface reconstruction approach.

抄録全体を表示

We propose a method of line-based Simultaneous Localization and Mapping (

) using non-overlapping multiple cameras for vehicles running in an urban environment. It uses corresponding line segments between images taken by different frames and different cameras. The contribution is a novel line segment matching algorithm by warping processing based on urban structures. This idea significantly improves the accuracy of line segment matching when viewing direction are very different, so that a number of correspondences between front-view and rear-view cameras can be found and the accuracy of can be improved. Additionally, to enhance the accuracy of we apply a geometrical constraint of urban area for initial estimation of 3D mapping of line segments and optimization by bundle adjustment. We can further improve the accuracy of by combining points and lines. The position error is stable within 1.5m for the entire image dataset evaluated in this paper. The estimation accuracy of our method is as high as that of ground truth captured by RTK-GPS. Our high accuracy algorithm can be apply for generating a road map represented by line segments. According to an evaluation of our generating map, true positive rate around the vehicle exceeding 70% is achieved.

抄録全体を表示

Service robots gain both geometric and semantic information about the environment with the help of semantic mapping, providing more intelligent services. However, a majority of studies for semantic mapping thus far require priori knowledge 3D object models or maps with a few object categories that neglect separate individual objects. In view of these problems, an object-oriented 3D semantic mapping method is proposed by combining state-of-the-art deep-learning-based instance segmentation and a visual simultaneous localization and mapping (

) algorithm, which helps robots not only gain navigation-oriented geometric information about the surrounding environment, but also obtain individually-oriented attribute and location information about the objects. Meanwhile, an object recognition and target association algorithm applied to continuous image frames is proposed by combining visual , which uses visual consistency between image frames to promote the result of object matching and recognition over continuous image frames, and improve the object recognition accuracy. Finally, a 3D semantic mapping system is implemented based on Mask R-CNN and ORB-2 frameworks. A simulation experiment is carried out on the ICL-NUIM dataset and the experimental results show that the system can generally recognize all the types of objects in the scene and generate fine point cloud models of these objects, which verifies the effectiveness of our algorithm.

抄録全体を表示

The visual

system requires precise localization. To obtain consistent feature matching results, visual features acquired by neural networks are being increasingly used to replace traditional manual features in situations with weak texture, motion blur, or repeated patterns. However, to improve the level of accuracy, most deep learning enhanced systems, which have a decreased efficiency. In this paper, we propose Coarse TRVO, a visual odometry system that uses deep learning for feature matching. The deep learning network uses a CNN and transformer structures to provide dense high-quality end-to-end matches for a pair of images, even under indistinctive settings with low-texture regions or repeating patterns occupying the majority of the field of view. Meanwhile, we made the proposed model compatible with NVIDIA TensorRT runtime to boost the performance of the algorithm. After obtaining the matching point pairs, the camera pose is solved in an optimized way by minimizing the re-projection error of the feature points. Experiments based on multiple data sets and real environments show that Coarse TRVO achieves a higher robustness and relative positioning accuracy in comparison with the current mainstream visual system.

抄録全体を表示

　 In order to evaluate the applicability of 3D scanners for field survey on surface ruptures, we examined the scanning accuracy, point cloud density, usability, and time efficiency of the instruments of three different

methods, Avia for LiDAR , ZED 2 for Visual , and iPad Pro for Depth

　 We conducted experimental surveys on the surface ruptures associated with the 2016 Kumamoto Earthquake at two locations. One is the surface rupture preserved as the earthquake heritage in the Aso field of Tokai University, while another is a normal fault rupture in the forested area at Miyayama, Nishihara Village, Kumamoto Prefecture. All the scanners obtained detailed point clouds, from which we successfully made digital surface models, cross-profiles and contour maps in a few tens of minutes. We came to know that Avia is most effective among the three scanners for wide-area mapping and that iPad Pro is a useful handy instrument for mapping limited areas. From our experimental survey, it is highly recommended to use Avia and iPad Pro together (in the field) in order to collect detailed geometric data of surface ruptures immediately after earthquake.

抄録全体を表示