计算机科学
人工智能
计算机视觉
机械手
机械手
感知
机器人
机器人学
机械臂
触觉传感器
触觉知觉
人机交互
工程类
作者
Wenhui Yang,Hongliang Zhao,Chengxiao He,Xiaowei Shi,Longhui Qin
出处
期刊:IEEE-ASME Transactions on Mechatronics
[Institute of Electrical and Electronics Engineers]
日期:2026-08-01
卷期号:31 (4): 4307-4318
被引量:1
标识
DOI:10.1109/tmech.2026.3719749
摘要
Robotic hands capable of sensitive tactile perception are expected in contact-rich scenarios to enable dexterous manipulations, whereas often suffering from high complexity and cost. In this article, we present a three-fingered soft-rigid-hybrid hand with embedded tactile perception capability, called PAM-driven soft-rigid-hybrid 3-fingered tactile hand (PDSR3-Tac hand), which is driven by pneumatic artificial muscle. Fully mimicking the human finger, each underactuated finger mainly consists of a rigid phalanx in the center, surrounding soft skin, and seven embedded tactile sensing elements (SEs) that are sensitive to both static and dynamic stimuli. Three such fingers simulating the thumb, index, and middle finger, constitute a whole hand in combination with a palm, a forearm, and pneumatic control system. After establishment of the kinematic model, an object recognition algorithm is proposed to realize one-grasp enabled tactile recognition via the fusion of total 21 channels of tactile SEs. Finally, the kinematic and perception performance of PDSR3-Tac hand is verified, and a contact-aware trajectory correction task is realized leveraging the full-hand tactile perception. In the application of tactile recognition, the accuracy exceeds 99.5% in both material and object identification with only one grasp, which further validates its performance and provides insights for more dexterous manipulations.
科研通智能强力驱动
Strongly Powered by AbleSci AI