| Budget Amount *help |
¥1,400,000 (Direct Cost: ¥1,400,000)
Fiscal Year 2005: ¥700,000 (Direct Cost: ¥700,000)
Fiscal Year 2004: ¥700,000 (Direct Cost: ¥700,000)
|
|---|
| Research Abstract |
In this research, two types of model for the motion of human arms were constructed. One is a static model to describe human-like pose of arms while human puts his hand on a desk. The other is a dynamic model to describe position and velocity of the hand while human is moving his arms. DH-method, which is often used in a field of Robot engineering, was adopted to describe the arm's pose. Relation between angles of shoulder, elbow, and wrist, length of the arms, and position of the hand was estimated by observed data of several positions of arms. Two types of linear models depending on the hand position were obtained. Dominant two regions of the hand position were divided by fuzzy clustering method, c-means and it was showed that these regions were corresponding to the region where human can reach his hand easily or uneasily through some experiments. Unified model of these two models was constructed as fuzzy model in which membership functions of the hand position belonging to the region were estimated by neural network. This model was verified by some experiences.
Based on above results, dynamic model for iterative motion of the arms in one region, where human can move his hand easily, was estimated by vector autoregressive model analysis. Estimated vector autoregressive model showed relation between angles and gave a rate of contribution to each other angles and described the relation of these angles for practiced motion of arms.
|
