Effect of polymer density on polyethylene hollow fiber membrane formation via thermally induced phase separation

Microporous high‐density polyethylene (HDPE) and low‐density polyethylene (LDPE) hollow fiber membranes were prepared from polyethylene–diisodecyl phthalate solution via thermally induced phase separation. Effect of the polyethylene density on the membrane structure and performance was investigated. The HDPE membrane showed about five times higher water permeability than the LDPE membrane because it had the larger pore and the higher porosity at the outer membrane surface. The formation of the larger pore was owing to both the initial larger structure formed by spinodal decomposition and the suppression of the diluent evaporation from the outer membrane surface due to the higher solution viscosity. © 2004 Wiley Periodicals, Inc. J Appl Polym Sci 93: 471–474, 2004     

Self-generation of reward by logarithmic transformation of multiple sensor evaluations

Artificial Life and Robotics - Although the design of the reward function in reinforcement learning is important, it is difficult to design a system that can adapt to a variety of environments and...     

Reusing Primitive and Acquired Motion Knowledge for Gait Generation of a Six-legged Robot Using Genetic Programming

There has been growing interest in motion planning problems for mobile robots. In this field, the main research is to generate a motion for a specific robot and task without previously acquired motions. However it is too wasteful not to use hard-earned acquired motions for other tasks. Here, we focus on a mechanism of reusing acquired motion knowledge and study a motion planning system able to generate and reuse motion knowledge. In this paper, we adopt a tree-based representation for expressing knowledge of motion, and propose a hierarchical knowledge for realizing a reuse mechanism. We construct a motion planning system using hierarchical knowledge as motion knowledge and using genetic programming as a learning method. We apply a proposed method for the gait generation task of a six-legged locomotion robot and show its availability with computer simulation.     

Solidification of Al2O3–YAG eutectic composites with off-metastable eutectic composition from undercooled melt produced by melting Al2O3–YAP eutectics

It has been reported that solidification of the Al₂O₃–YAG equilibrium eutectic structure follows melting of the Al₂O₃–YAP metastable eutectic structure. Since the exothermic heat due to solidification was consumed by the endothermic heat due to melting, a fine and uniform eutectic structure was obtained. However, the composition of the Al₂O₃–YAG eutectic structure is restricted to the metastable eutectic composition. In this paper, Al₂O₃–YAG eutectic compacts with an off-metastable eutectic composition were prepared by the addition of Al₂O₃ particles to Al₂O₃–YAP eutectic particles with diameters less than 20 μm. In compositions ranging from 18.5 mol%Y₂O₃ to 13.5 mol%Y₂O₃, dense Al₂O₃–YAG eutectic compacts were formed without any Al₂O₃ segregation. The flexural strength and the fracture toughness remained almost unchanged with the increase in the Al₂O₃ phase. The addition of Al₂O₃ particles to the Al₂O₃–YAP eutectic particles enabled the matrix phase to change from the YAG phase to the Al₂O₃ phase.     

Synthesis of Microcrystalline Hematite and Magnetite in Organic Solvents and Effect of a Small Amount of Water in the Solvents

Microcrystalline magnetite (Fe₃O₄) particles having diameters of 7-34 nm were synthesized by thermal treatment of iron(III) acetylacetonate in organic solvents at high temperatures (423-573 K) and under autogenous pressures of the solvents. When the treatment was conducted in 1-propanol containing a small amount of water (1-3 vol%), nanosized hematite (α-Fe₂O₃) crystallized, which then was reduced to magnetite after a prolonged reaction time, indicating that a small amount of water in 1-propanol drastically changed the reaction mechanism for the formation of iron oxides.     

Reduction of state space in reinforcement learning by sensor selection

Much research has been conducted on the application of reinforcement learning to robots. Learning time is a matter of concern in reinforcement learning. In reinforcement learning, information from sensors is projected on to a state space. A robot learns the correspondence between each state and action in state space and determines the best correspondence. When the state space is expanded according to the number of sensors, the number of correspondences learnt by the robot is increased. Therefore, learning the best correspondence becomes time consuming. In this study, we focus on the importance of sensors for a robot to perform a particular task. The sensors that are applicable to a task differ for different tasks. A robot does not need to use all installed sensors to perform a task. The state space should consist of only those sensors that are essential to a task. Using such a state space consisting of only important sensors, a robot can learn correspondences faster than in the case of a state space consisting of all installed sensors. Therefore, in this paper, we propose a relatively fast learning system in which a robot can autonomously select those sensors that are essential to a task and a state space for only such important sensors is constructed. We define the measure of importance of a sensor for a task. The measure is the coefficient of correlation between the value of each sensor and reward in reinforcement learning. A robot determines the importance of sensors based on this correlation. Consequently, the state space is reduced based on the importance of sensors. Thus, the robot can efficiently learn correspondences owing to the reduced state space. We confirm the effectiveness of our proposed system through a simulation.     

Reinforcement learning for quasi-passive dynamic walking of an unstable biped robot

A class of biped locomotion called Passive Dynamic Walking (PDW) has been recognized to be efficient in energy consumption and a key to understand human walking. Although PDW is sensitive to the initial condition and disturbances, studies of Quasi-PDW which incorporates supplemental actuators have been reported to overcome this sensitivity. In this article, we propose a reinforcement learning method designed particularly for Quasi-PDW of a biped robot whose possession of knees makes the system unstable. Simulations show that the learning is quickly accomplished after 1000 episodes, and the obtained controller is robust against variations in the slope gradient and sudden perturbations.     

Prediction of space weather by adaptive information processing

Space weather can be predicted using data from satellites. For example, the condition of high-energy electrons at geostationary orbit is vital in providing warnings for spacecraft operations. We investigate an adaptive predictor based on intelligent information processing. The predictor forecasts the condition of high-energy electrons at geostationary orbit 24 h ahead. This article focuses on adaptation and performance of the predictor. Our proposed model succeeded in forecasting the high-energy electron flux at geostationary orbit 24 h ahead. Furthermore, we also consider the relationship between the prediction performance and tuning parameters.     

Crystallization and Transformation of Aluminum Orthophosphates in Organic Solvent Containing a Small Amount of Water

Aluminum orthophosphate (AlPO₄) samples were synthesized by high-temperature (200-300°C) hydrolysis of a mixture of aluminum triisopropoxide and triethyl phosphate with a small amount of water dissolved in an organic solvent from the gas phase. At 250°C, cristobalite-type AlPO₄, which is the metastable phase at that temperature, was first crystallized from the amorphous phase and then transformed to stable quartz-type AlPO₄ (berlinite) with a prolonged reaction time. When the amount of water charged into the system was increased, berlinite was formed immediately. The formation of the metastable phase could be attributed to the low solubility of AlPO₄ in the organic solvent containing a small amount of water.     

Promotion effect of bifunctional five‐membered cyclic dithiocarbonate on curing of one‐component epoxy resin by imines as latent initiator

This article deals with curing of a one‐component epoxy resin containing a bifunctional five‐membered cyclic dithiocarbonate 1 with an imine 2 as a latent initiator. When 1 was added to a mixture of epoxy resin and 2 , the curing rate and initial adhesive strength increased higher than that without 1 . It was supposed that the reaction of 1 and an amine released from 2 was much faster than the amine–epoxide reaction, and a thiol group formed by the reaction of 1 with the amine accelerated the amine–epoxide reaction. The adhesive strength of the one‐component epoxy resin containing 10 mol % of 1 exhibited the highest value. © 2004 Wiley Periodicals, Inc. J Appl Polym Sci 94: 961–964, 2004