計測自動制御学会論文集
Online ISSN : 1883-8189
Print ISSN : 0453-4654
ISSN-L : 0453-4654
45 巻, 1 号
選択された号の論文の9件中1~9を表示しています
論文
計測
  • 山口 崇, 加嶋 良年, 笹田 一郎
    2009 年 45 巻 1 号 p. 1-10
    発行日: 2009年
    公開日: 2011/11/01
    ジャーナル フリー
    In this paper, a fast method of estimating the position of a moving object for a magnetic motion capture system is presented. The system uses two quasi-uniform magnetic fields along the X-axis and the Z-axis and one gradient magnetic field having a null point at the origin. Each of these magnetic fields is produced sequentially by using two pairs of square coils wound along four edges of a cube where one pair is rotated through 90 degree from the other. A triaxial magnetometer is attached to the moving object to measure these magnetic field vectors. The position of the object is calculated from three scalar products among the measured magnetic field vectors. The calculation is done approximately by using a set of truncated Taylor serieses, where Wynn's vector epsilon-algorithm is used together to expand the area of convergence numerically. This method of calculation contains no iterative process and is much faster than the Gauss-Newton method. The position of the object is stably estimated by this method within a cubic area having an edge length of 60 percent of the side length of the coils. A prototype system using a triaxial Hall magnetometer was build. The experimental results show that the method works properly in a real system.
  • 栗原 陽介, 渡辺 嘉二郎, 小林 一行, 田中 博
    2009 年 45 巻 1 号 p. 11-17
    発行日: 2009年
    公開日: 2011/11/01
    ジャーナル フリー
    Sleep disorders disturb the recovery from mental and physical fatigues, one of the functions of the sleep. The majority of those who with the disorders are suffering from Sleep Apnea Syndrome (SAS). Continuous Hypoxia during sleep due to SAS cause Circulatory Disturbances, such as hypertension and ischemic heart disease, and Malfunction of Autonomic Nervous System, and other severe complications, often times bringing the suffers to death. In order to prevent these from happening, it is important to detect the SAS in its early stage by monitoring the daily respirations during sleep, and to provide appropriate treatments at medical institutions. In this paper, the Pneumatic Method to detect the Apnea period during sleep is proposed. Pneumatic method can measure heartbeat and respiration signal. Respiration signal can be considered as noise against heartbeat signal, and the decrease in the respiration signal due to Apnea increases the Average Mutual Information of heartbeat. The result of scaling analysis of the average mutual information is defined as threshold to detect the apnea period. The root mean square error between the lengths of Apnea measured by Strain Gauge using for reference and those measured by using the proposed method was 3.1 seconds. And, error of the number of apnea times judged by doctor and proposal method in OSAS patients was 3.3 times.
制御
システム・情報
  • 永吉 雅人, 村尾 元, 玉置 久
    2009 年 45 巻 1 号 p. 41-50
    発行日: 2009年
    公開日: 2011/11/01
    ジャーナル フリー
    This paper presents a technique for reinforcement learning to deal with both discrete and continuous state space systems in POMDPs while keeping the state space of an agent compact. First, in our computational model for MDP environments, a concept of “ state space filtering ” has been introduced and constructed to make the state space of an agent smaller properly by referring to “ entropy ” calculated based on the state-action mapping. The model is extended to be applicable in POMDP environments by introducing the mechanism of utilizing effectively of history information. The extended model is capable of being dealt with a continuous state space as well as a discrete state space by the extended model. Here, the mechanism of adjusting the amount of history information is also introduced so that the state space of an agent should be compact. Moreover, some computational experiments with a robot navigation problem with a continuous state space have been carried out. The potential and the effectiveness of the extended approach have been confirmed through these experiments.
  • 香川 高弘, 山科 秀貴, 宇野 洋二
    2009 年 45 巻 1 号 p. 51-59
    発行日: 2009年
    公開日: 2011/11/01
    ジャーナル フリー
    An efficient and stable gait control is an essential problem to develop a legged locomotor device for paraplegics. In this study, we investigate a necessary condition of the ballistic walking to avoid a backward balance loss. The condition derived by an inverted pendulum model is represented as a simple relationship between a position and velocity of a body center of mass at toe-off. The condition was validated through simulation experiments of a 7-link musculoskeletal model and gait measurement experiments of normal and paraplegic subjects. The results of the model simulation showed a good agreement with some predictions of the inverted pendulum model. The measured center of mass trajectories of normal and paraplegic gaits were satisfied with the necessary condition. These results suggest that the necessary condition is effective to avoid a backward falling during walking. In addition, energy input was required in a double support phase while the trajectory followed the ballistic movement of the inverted pendulum in a single support phase for a normal subject. These results suggest that a power assist control to be satisfied with the necessary condition during a double support phase and a ballistic gait generation during a single support phase are required for a paraplegic locomotor with efficiency and stability.
  • 池田 徹志, 石黒 浩, 西村 拓一
    2009 年 45 巻 1 号 p. 60-68
    発行日: 2009年
    公開日: 2011/11/01
    ジャーナル フリー
    To realize intelligent environments that understand and support our daily activities, it is important to observe positions and identities of people in the environment and many studies have been proposed using sensor networks. Floor sensors can reliably detect current positions of people and prevent invasion of privacy, but it is difficult to identify people. To solve the problem, we propose to integrate acceleration sensors that are attached to the human body. Since the signals from floor sensors and acceleration sensors are correlated when they observe the same person, these signals are not independent. The correlation between the signals is evaluated based on a statistical test to find correct association of positions to IDs. People tracking examples are shown to confirm the effectiveness of the proposed method. Significant improvement in correct association rate is achieved compared to the results using only floor sensors.
ショート・ペーパー
feedback
Top