Hirofumi Kanazaki

The University of Tokyo, Edo, Tōkyō, Japan

Are you Hirofumi Kanazaki?

Claim your profile

Publications (8)0 Total impact

  • [Show abstract] [Hide abstract]
    ABSTRACT: This paper proposes an alternative solution to a mapping problem in two different cases; when bearing measurements to features (landmarks) and odometry are measured and when bearing and range measurements to features are measured. Our approach named M-SEIFD (Mapping by Sequential Estimation of Inter-Feature Distances) first estimates inter-feature distances, then finds global position of all the features by enhanced multi-dimensional scaling (MDS). M-SEIFD is different from the conventional SLAM methods based on Bayesian filtering in that robot self-localization is not compulsory and that M-SEIFD is able to utilize prior information about relative distances among features directly. We show that M-SEIFD is able to achieve a decent map of features both in simulation and in real-world environment with a mobile robot.
  • Source
    Takehisa Yairi, Hirofumi Kanazaki
    [Show abstract] [Hide abstract]
    ABSTRACT: In this paper, we introduce an alternative solution to the bearing-only mapping problem in which a mobile robot builds a map of features (landmarks) using only relative bearing measurements to them and odometry information. Our approach named BOM-STMDS (bearing-only mapping by sequential triangulation and multi-dimensional scaling) first tries to estimate relative distances among the features, then finds two-dimensional coordinates of all features by using multi-dimensional scaling (MDS) and its enhancements. BOM- STMDS is different from the conventional BOSLAM methods based on Bayesian filtering in that robot self-localization is not mandatory. Another remarkable property is that BOM-STMDS is able to utilize prior information about relative distances among features efficiently. In the experiment, the performance of BOM-STMDS is shown to be competitive with a conventional EKF-based BOSLAM method.
    Robotics and Automation, 2008. ICRA 2008. IEEE International Conference on; 06/2008
  • [Show abstract] [Hide abstract]
    ABSTRACT: We propose a sequential variational Bayes method, which is a recursive formulation of variational Bayes method, extended for online learning. We derived a novel data association filtering method for multiple targets, named variational Bayes data association filter (VBDAF). To estimate multiple targets' states, data association is an important problem, when data don't have unique labels and we can only associate data and targets probabilistically. EM algorithms or variational Bayes methods have been used for estimation problems with missing values such as data labels, but they are batch formulations. JPDAF have been widely used for multiple targets tracking. It is an extended filtering method based on sequential Bayes methods such as Kalman Filter, and approximation in the sense of finite mixture distributions, where VBDAF is approximate in the sense of KL divergence. We demonstrate VBDAF, in application of online multiple target localization.
    Intelligent Sensors, Sensor Networks and Information, 2007. ISSNIP 2007. 3rd International Conference on; 01/2008
  • [Show abstract] [Hide abstract]
    ABSTRACT: This paper proposes an alternative solution to a mapping problem in two different cases; when bearing measurement to features (landmarks) and odometry are measured and when local position of features are measured. Our approach named M-SEIFD (Mapping by Sequential Estimation of Inter-Feature Distances) first estimates inter-feature distances, then finds global position of all features by enhanced multi-dimensional scaling (MDS). M-SEIFD is different from the conventional SLAM methods based on Bayesian filtering in that robot self-localization is not compulsory and that M-SEIFD is able to utilize prior information about relative distances among features directly. We show that M-SEIFD is able to achieve a decent map of features both in simulation and in real-world environment with a mobile robot.
    PRICAI 2008: Trends in Artificial Intelligence, 10th Pacific Rim International Conference on Artificial Intelligence, Hanoi, Vietnam, December 15-19, 2008. Proceedings; 01/2008
  • [Show abstract] [Hide abstract]
    ABSTRACT: In this paper, we describe a formulation and EM method to fuse heterogeneous sensor data for localizing and identifying objects. The heterogeneous sensors are such as cameras, RFID readers and range finders. They are different from each other in degree of accuracy of localizing and identifying objects. Errors in identifying objects is a problem which makes localization difficult. We formulate a localization and identification problem with a hidden variable R which describes associations between objects and observed values. We estimate objects' positions and the association vector R by an EM algorithm. We illustrate the algorithm in a case of using simulated cameras and RFID readers. Experimental results show that our method can localize and identify objects more accurately than other methods without the R
    SICE-ICASE, 2006. International Joint Conference; 11/2006
  • [Show abstract] [Hide abstract]
    ABSTRACT: Position estimation and tracking of multiple objects by vision sensors is one of the most fundamental technologies. While the vision sensors provide high accuracy measurements for position estimation, they require suitable features of objects for accurate recognition and detection as prior knowledge. Especially, learning of appearance based features of objects requires large quantities of training data, which makes development costs. This paper proposes a method for learning appearance based features of objects using auxiliary data of RFID. In this method, the RFID device is used as a supervisor to semi-automatically construct the training data set for each object. Since it is difficult to observe what ID does an object image correspond to, this problem comes down to supervised learning using incompletely labeled features. This paper proposes a learning method using Kernel PCA and EM algorithm, and verifies the effectiveness and robustness of this method
    SICE-ICASE, 2006. International Joint Conference; 11/2006
  • [Show abstract] [Hide abstract]
    ABSTRACT: This research intends to build up the probabilistic models of three heterogeneous sensors, camera, range sensor and RFID for dealing with objects localization problems, and also aims to fulfil the sensor fusion with these models. In this paper, we describe the mathematical significance for the sensor models, and define concrete models. Then we show the effectiveness of sensor fusion with these models from the simulation
    SICE-ICASE, 2006. International Joint Conference; 11/2006
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: We apply a variational approximation for multiple-target localiza-tion, and propose Variational Approximation Data Association Filter(VADAF) method, which minimize KL divergence between marginalized likelihood and approximated one. For multiple-target localization, we have to solve data association problem. The data association problem is that we can not associate data and targets deterministically, when data don't have unique labels associated to targets. JPDAF is widely used for multiple-target tracking (MTT). It is extended filtering method based on Sequential Bayesian Esti-mation methods, such as Kalman Filter. Our method is not only based on the sequential bayes estimation, but based on variational approximation method. Our main contribution is derivation of variational approxi-mated likelihood of targets' states, and optimize it by minimizing KL divergence. It is more precisely than mixture likelihood of JPDAF method.