Study on Shape-Classification of Prism Objects by Soft Tactile Sensor Array System with Contact Strength

摘要

We have developed a soft tactile sensor array system that is composed of five links. The surface of the links is covered with soft tactile sensor that is made of semi-circular silicone rubber with 235 on-off switches. When the links grasp an object, we get on-off data from them, and transform the data into a time-series data matrix form. We extract 8 neighborhood on-sell patterns from the time-series data matrix, and count their patterns. Then we make a vector from the frequency of these patterns. We categorize these vectors by Mahalanobis' generalized distance. We have classified prism objects by using soft tactile sensor array system with contact strength information. Here we indicate the result and the methods of using their contact strength.