Automatic Identification of Participants in Haptic Systems

Abstract

Biometric systems identify of users based on behavioral or physiological characteristics. This paper explores the feasibility of automatically identifying participants in haptic systems. Such a biometric system would lead to important and interesting applications such as continuous authentication in tele-operation. In order to test this feasibility, we designed a haptic system in which position, velocity, force and torque data from the tool was continuously measured and stored. Using this system, users navigated a simple maze where the user generates a continuous path from start to finish. Subsequently, several algorithms were developed to extract characteristic biometric features from the measured data. A 78.8% probability of verification was observed for data from trained users. Overall, the paper suggests the possibility of extracting identity information in a real world haptic system