Novel application of 3D range image sensor for personal identification based on skeletal information

Abstract

Robots that perform services for humans are expected to be effective and useful. Guide robots are examples of such service robots. For a guide robot to act according to the needs of the guidance target, the guide robot must have the ability to differentiate between the guidance target and other individuals. Therefore, the guide robot requires personal identification functionality. Here, a method to acquire an individual's upper-body skeletal information using a three-dimensional range image sensor, and to perform personal identification based on these data, is proposed. Personal identification using the proposed method is performed for 13 subjects, and identification is achieved with an accuracy of 98.4%. These results indicate the feasibility of personal identification using upper-body skeletal information.