日本ロボット学会誌
Online ISSN : 1884-7145
Print ISSN : 0289-1824
ISSN-L : 0289-1824
29 巻 , 1 号
選択された号の論文の19件中1~19を表示しています
特集について
解説
事例紹介
談話
実用技術紹介
論文
  • 市川 泰久, 望山 洋, 藤本 英雄
    2011 年 29 巻 1 号 p. 47-54
    発行日: 2011年
    公開日: 2011/02/25
    ジャーナル フリー
    In this paper, we propose an impulse force generator for catapulting an object. The proposed impulse force generator utilizes snap-through buckling of an elastic body. The distinguished feature of the device is to adopt a serial chain of torsion springs with both high elasticity and flexibility as its elastic body. The use of the serial chain of torsion springs enables us to increase the energy density of the elastic body. Experimental results show that the proposed compact and lightweight impulse force generator has high capability of catapulting an object. The generated maximum momentum is of 16[Ns] in spite that the maximum driving torque necessary for snap-through buckling is only of 21[Nm].
  • 山口 明彦, 高松 淳, 小笠原 司
    2011 年 29 巻 1 号 p. 55-66
    発行日: 2011年
    公開日: 2011/02/25
    ジャーナル フリー
    Discrete action sets are often used in many reinforcement learning (RL) applications in robot control, since such sets are compatible with many RL methods and sophisticated architectures, such as Q(λ)-learning [1] and the Dyna. However, one of the problems is the absence of general principles on designing a discrete action set for robot control in higher dimensional input space. In this paper, we propose a discrete action set DCOB which is generated from the given basis functions (BFs) for approximating a value function. Though the DCOB is a discrete set, it has an ability to acquire high performance. Moreover, we utilize a method that generates a set of BFs based on the dynamics of the robot to reduce the number of the BFs. This way also makes the DCOB compact. Thus, the DCOB is compact and has an ability to acquire performance. Moreover, we also propose a method WF-DCOB, where the wire-fitting [2] is utilized to learn within a continuous action space which the DCOB discretizes. The purpose of the WF-DCOB is to evaluate the possibility of acquiring higher performance. Our proposition in the WF-DCOB is to constrain and initialize the parameters to relax the instability of the wire-fitting. We apply the proposed methods for a humanoid robot to learn crawling motion. The simulation results demonstrate outstanding advantages of the proposed method both in learning speed and ability to acquire performance, compared to conventional action space.
  • 小川原 光一, 田邉 康史, 倉爪 亮, 長谷川 勉
    2011 年 29 巻 1 号 p. 67-76
    発行日: 2011年
    公開日: 2011/02/25
    ジャーナル フリー
    Frequent patterns in time series data are useful clues to learn previously unknown events in an unsupervised way. In this paper, we propose a method for detecting frequent patterns in long time series data efficiently. The major contribution of the paper is two-fold: (1) Partly Locality Sensitive Hashing (PLSH) is proposed to find frequent patterns efficiently and (2) the problem of finding consecutive time frames that have a large number of frequent patterns is formulated as a combinatorial optimization problem which is solved via Dynamic Programming (DP) in polynomial time O (N 1+1/α) thanks to PLSH where N is the total amount of data. The proposed method was evaluated by detecting frequent whole body motions in a video sequence as well as by detecting frequent everyday manipulation tasks in motion capture data.
  • 星野 由紀子, 河本 献太, 野田 邦昭, 佐部 浩太郎
    2011 年 29 巻 1 号 p. 77-88
    発行日: 2011年
    公開日: 2011/02/25
    ジャーナル フリー
    Continual and autonomous learning are key features for a developmental agent in open-ended environments. This paper presents a mechanism of self-regulated learning to realize them. Considering the fact that learning progresses only when the learner is exposed to appropriate level of uncertainty, we propose that an agent's learning process be guided by the following two metacognitive strategies throughout its development: (a) Switch of behavioral strategies to regulate the level of expected uncertainty, and (b) Switch of learning strategies in accordance with the current subjective uncertainty. With this mechanism, we demonstrate efficient and stable online learning of a maze where only local perception is provided: the agent autonomously explores an environment of significant-scale, and self-develops an internal model that properly describes the hidden structure behind its experience.
  • 田原 健二, 有本 卓, 吉田 守夫
    2011 年 29 巻 1 号 p. 89-98
    発行日: 2011年
    公開日: 2011/02/25
    ジャーナル フリー
    This paper proposes a novel object manipulation method to regulate the position and attitude of a rigid object of parallelepiped with maintaining stable grasping by a triple soft-fingered hand without use of any external sensing. In the authors' previous works, “the Blind Grasping” control scheme was proposed for a pair of soft-fingered hand to accomplish dynamic object grasping without use of any external sensing. However, it has just achieved a stable object grasping, and a simultaneous object position and attitude control has not yet been treated, so far. To this end, a ternary finger in addition to a pair of fingers is introduced in order to enhance the Blind Grasping manner to “the Blind Manipulation” manner. Since the Blind Grasping control input has been designed for a pair of fingers, in this paper it is renewed to install it into the triple-fingered robotic hand. Moreover, both the virtual object position and attitude are utilized by defining a virtual frame instead of real information of the object position and attitude in order to achieve the Blind Manipulation. First, the overall dynamics, which considers rolling contact constraints between each fingertip and object surface, is derived. Second, a control signal to regulate the virtual object position and attitude without use of any external sensing, “the Blind Manipulation”, is designed. Next, the stability of the system is discussed briefly, and finally the usefulness of our proposed manipulation method is demonstrated through a numerical simulation.
  • 浅野 文彦
    2011 年 29 巻 1 号 p. 99-110
    発行日: 2011年
    公開日: 2011/02/25
    ジャーナル フリー
    This paper proposes a novel method for generating a dynamic gait based on anterior-posterior asymmetric impact posture tilting the robot's center of mass forward. The primary purpose of this method is to asymmetrize the impact posture by actuating the robot's telescopic-legs to make overcoming the potential barrier at mid-stance easy, and the mechanical energy is accordingly restored. First, we introduce a planar rimless wheel model with telescopic legs, and investigate the validity of the stance-leg extension control. The basic properties and efficiency of the generated gait are also numerically analyzed. Second, we extend the method to a planar telescopic-legged biped model, and investigate the validity through numerical simulations. Furthermore, we discuss the role of asymmetric shape of human foot from the brake effect point of view through efficiency analysis taking the ankle-joint actuation into account.
  • 西脇 光一, 加賀美 聡
    2011 年 29 巻 1 号 p. 111-121
    発行日: 2011年
    公開日: 2011/02/25
    ジャーナル フリー
    This paper presents a novel robust framework for online walking control that can handle unknown terrain. The actual motion of the robot in the absolute coordinate system is estimated and used as the initial conditions of short cycle online walking pattern generation in the framework so that the walking pattern generation is put into the feedback loop of the balance maintenance of the actual walking. We used attitude sensor system for estimating the absolute motion, and the walking patterns are generated at 20[ms] cycle by using full body dynamic model. Since the initial conditions are decided from the estimated actual posture, the pattern generated in each cycle is discontinuous to the one generated in the previous cycle. Therefore, a local sensor feedback controller is developed that can execute discontinuous segments of the trajectory and control the ground reaction force. Damping control of the foot position by using position-controlled leg joints is adopted to attain the desired ground reaction force. In addition, damping control of the torso inclination is implemented in the local sensor feedback to suppress the divergence of the torso motion from that commanded in the absolute coordinate system. The proposed framework is implemented on the full-size humanoid HRP-2, and validated through the experiments including walking on many types of unknown terrain with up to 10 degrees slope uncertainty and over 20[mm] in height uncertainty.
feedback
Top