TY - JOUR
T1 - Deep vibro-tactile perception for simultaneous texture identification, slip detection, and speed estimation
AU - Massalim, Yerkebulan
AU - Kappassov, Zhanat
AU - Varol, Huseyin Atakan
N1 - Funding Information:
Funding: This work was partially supported by the NU Faculty-development competitive research grants program “Variable Stiffness Tactile Sensor for Robot Manipulation and Object Exploration” 110119FD45119 and the Ministry of Education and Science of the Republic of Kazakhstan grant “Methods for Safe Human Robot Interaction with Variable Impedance Actuated Robots”.
Publisher Copyright:
© 2020 by the authors Licensee MDPI, Basel, Switzerland.
PY - 2020/8/1
Y1 - 2020/8/1
N2 - Autonomous dexterous manipulation relies on the ability to recognize an object and detect its slippage. Dynamic tactile signals are important for object recognition and slip detection. An object can be identified based on the acquired signals generated at contact points during tactile interaction. The use of vibrotactile sensors can increase the accuracy of texture recognition and preempt the slippage of a grasped object. In this work, we present a Deep Learning (DL) based method for the simultaneous texture recognition and slip detection. The method detects non-slip and slip events, the velocity, and discriminate textures—all within 17 ms. We evaluate the method for three objects grasped using an industrial gripper with accelerometers installed on its fingertips. A comparative analysis of convolutional neural networks (CNNs), feed-forward neural networks, and long short-term memory networks confirmed that deep CNNs have a higher generalization accuracy. We also evaluated the performance of the highest accuracy method for different signal bandwidths, which showed that a bandwidth of 125 Hz is enough to classify textures with 80% accuracy.
AB - Autonomous dexterous manipulation relies on the ability to recognize an object and detect its slippage. Dynamic tactile signals are important for object recognition and slip detection. An object can be identified based on the acquired signals generated at contact points during tactile interaction. The use of vibrotactile sensors can increase the accuracy of texture recognition and preempt the slippage of a grasped object. In this work, we present a Deep Learning (DL) based method for the simultaneous texture recognition and slip detection. The method detects non-slip and slip events, the velocity, and discriminate textures—all within 17 ms. We evaluate the method for three objects grasped using an industrial gripper with accelerometers installed on its fingertips. A comparative analysis of convolutional neural networks (CNNs), feed-forward neural networks, and long short-term memory networks confirmed that deep CNNs have a higher generalization accuracy. We also evaluated the performance of the highest accuracy method for different signal bandwidths, which showed that a bandwidth of 125 Hz is enough to classify textures with 80% accuracy.
KW - Accelerometers
KW - Convolutional neural networks
KW - Deep learning
KW - Long short-term memory
KW - Slip detection
KW - Tactile sensing
KW - Texture identification
UR - http://www.scopus.com/inward/record.url?scp=85088680099&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85088680099&partnerID=8YFLogxK
U2 - 10.3390/s20154121
DO - 10.3390/s20154121
M3 - Article
C2 - 32722353
AN - SCOPUS:85088680099
SN - 1424-8220
VL - 20
SP - 1
EP - 15
JO - Sensors (Switzerland)
JF - Sensors (Switzerland)
IS - 15
M1 - 4121
ER -