Studies of a Material Recognition System with an Artificial Finger Using a Pneumatic Tendon-Driven Mechanism

抄録

　Personal care robots are expected to be valuable labor force in Japan that is entering an aging society. When using the personal care robot in the same space as humans, there are many safety issues. Personal care robots are actively researched, and flexible pneumatic actuators are used there. Collecting environmental information is necessary for robots and humans to coexist safely. Research on the tactile field of the robot does not reproduce the human skin sensation and the like but utilizes data obtained from contact sensors and the like as a physical quantity different from human skin sensation. However, there are many problems that have not yet been elucidated for the mechanism of human material recognition. Tactile sensation is a sense that it is possible to detect the hardness, surface roughness, temperature of an object for the first time by touching and moving an object unlike vision and hearing, and it is a sensation that it is important to actively move a finger.

In this study, artificial fingers with joints for detecting the surface condition, hardness and temperature are fabricated using a strain gauge and a thermocouple, and by introducing active tactile perception using a pneumatic actuator, Develop pneumatic artificial fingers with near tactile sensory function. To make the texture factor equivalent to that obtained from humans by these, a material recognition system capable of discriminating the material equivalent to human by optimizing a neural network that simulates the connection structure of nerve cells in human brain.