We propose an interactive genetic algorithm (IGA) that evaluates individuals from among many people. This technique is effective for consensus building and in bringing together many opinions for applications such as enabling multiple people to design clothes together. With an IGA in which multiple people participate, a simple and easy evaluation interface is required. Our interface technique determines the solution group of the IGA using the tournament type evaluation. In this technique, two solution candidates are presented to evaluators who votes for the best solution candidate. Several evaluators judge the solution candidate's superiority or inferiority, thus deciding the relative rank of the two solutions. Evaluation points for all solution candidates are assigned depending on the relative value of the rank of the two solutions, which is based in turn on the results of the tournament competition. The solution obtained reflects the opinion of all evaluators accurately. In this study, the effectiveness of the proposed technique is verified using a simulation that employs multiple numerical evaluation agents instead of human evaluators. For each tournament competition, the results saturate when 60% of all evaluation agents participate in a vote; that is, the result is unaffected even if more evaluation agents participate. This means that the evaluator load can be reduced by 40%. Moreover, we compare the proposed technique with a general voting method that uses several evaluators voting for the first good individual from among all solution candidates in a displayed list. We confirm that the proposed technique is more effective than the general voting method, because many participants are satisfied.

View full abstract

We examined attentional attractiveness of visual cues such as used in presentation graphics. In experiments, we used four different types of visual cues and a visual search task on a computer display. In Experiment 1, it was revealed that two cues (flashing cue and zooming cue) were useful for reducing reaction time to find a target. In Experiment 2, we informed low probability of cue validity to observers. The results showed that the attentional attractiveness found in Experiment 1 could not explained by top-down control of visual attention. Applying these cues to information display would be expected to effectively guide people's attention to important information.

View full abstract

This paper describes facial muscular activities in drowsy expression. We measured facial images and facial electromyograms (F-EMGs) in a tracking task conducted with 17 participants. We investigated the mechanism of drowsy expression through analysis of the F-EMGs. Contraction of the frontalis muscle and relaxation of the zygomaticus major muscle were found to be the two common features of muscular activities in drowsy expression. The former muscular activity appeared in a medium drowsy condition, but not in arousal or dozing states. This result indicates a driver's resistance against drowsiness before reaching a dozing state, in that contraction of the frontalis muscle causes the driver's eyebrows to rise. The latter muscular activity became more obvious with an increase in the driver's drowsiness. This result indicates a decline in the driver's physical activity, in that relaxation of the zygomaticus major muscle causes the driver's cheeks to slack. Detecting these two features is considered to be useful in estimating a driver's drowsiness level by means of facial image processing. The activities of the orbicularis oris muscle, mentalis muscle, and masseter muscle, which causes the driver's “lips part (slightly open mouth)”, showed a variety of tendencies in respect of the drowsiness levels among different individuals. Consideration of the difference in muscular activities among individuals is necessary when the “lips part” feature is applied to estimation of a driver's drowsiness level through facial image processing.

View full abstract

This paper is looking at the ways of incorporating engineering emotional values into product design through Kansei engineering. The first part of the paper describes the Kansei engineering methodology. The second attempts to link emotions and product properties through various methods from Kansei engineering. Some results from the second part have been used in the first part as illustrated examples. The second part is used to validate Kansei engineering methodology through this study some questions have been investigated: Which methods within Kansei engineering can be deployed in design practices? Is Kansei engineering a tool which can be used by most designers to link emotions and product properties? Do users know how to describe their ideal products through words? And what is the future for Kansei engineering in India?

View full abstract

This paper proposes an offline sound volume feedback algorithm for a violin-playing robot and shows the validity of the algorithm. We built a 7 D.O.F. violin-playing robot and investigated the relationship between bow speed and sound volume. We found from experiments that the sound volume can be approximated by a third-order polynomial of the bow speed. Based on this relationship, we proposed an algorithm for sound volume feedback. We implemented the proposed algorithm onto the robot to clarify its effects. The result confirmed the validity of the proposed algorithm.

View full abstract