Biomimetic Real-Time Multimodal Tactile Perception and Haptics for Telepresence Humanoids

Abstract

Touch is critical to human object manipulation, enabling rapid optimal material grasping. Contrarily, telepresence robots seldom display fine sensory information to their operators, resulting in awkward object interaction. This paper details the biomimetic real-time multimodal haptics system of telepresence humanoid Avatar-Hubo and its tactual perception techniques. Inspired by human sensing, the proposed system uses a custom robotic hand and dynamic sensor to measure finger compliance and vibrations during object contact. Then, the proposed Real-Time Haptics Method processes sensor signals into metrics to quantify and identify material compliance and texture. Finally, a commercial haptic interface provides operators with appropriate cutaneous force and texture feedback. Experimentation with the Real-Time Haptics Method reveals over 80 % accuracy in recognizing smooth and rough items without requiring the contact force/velocity control used in similar systems. Hence, the proposed method enables telepresence robots like Avatar-Hubo to identify tactile properties and provide operators with rich tactile feedback in real-time.