Target Classification Method of Tactile Perception Data with Deep Learning

Abstract

In order to improve the accuracy of manipulator operation, it is necessary to install a tactile sensor on the manipulator to obtain tactile information and accurately classify a target. However, with the increase in the uncertainty and complexity of tactile sensing data characteristics, and the continuous development of tactile sensors, typical machine-learning algorithms often cannot solve the problem of target classification of pure tactile data. Here, we propose a new model by combining a convolutional neural network and a residual network, named ResNet10-v1. We optimized the convolutional kernel, hyperparameters, and loss function of the model, and further improved the accuracy of target classification through the K-means clustering method. We verified the feasibility and effectiveness of the proposed method through a large number of experiments. We expect to further improve the generalization ability of this method and provide an important reference for the research in the field of tactile perception classification.

Gradient adaptive sampling and multiple temporal scale 3D CNNs for tactile object recognition

“…As shown in Table 4 , the top 1 score of the proposed method surpasses the Sundaram et al ( 2019 ), Wang et al ( 2021 ), Zhang et al ( 2021 ), and Sharma ( 2022 ) by 16.43, 16.81, 8.72, and 6.99%, which indicates that our method has the highest classification accuracy on the MIT-STAG dataset. Meanwhile, the proposed method of KC surpasses the Sundaram et al ( 2019 ), Wang et al ( 2021 ), Zhang et al ( 2021 ), and Sharma ( 2022 ) by 17.18, 17.57, 9.22, and 7.4%, which indicates that our method has the lowest degree of confusion which can also be seen in Figure 4 .…”

“… Comparisons with state-of-the-art models in terms of confusion matrix on the MIT-STAG dataset. (A) Sundaram et al, 2019 , (B) Wang et al, 2021 , (C) Zhang et al, 2021 , (D) Sharma, 2022 , (E) Ours. …”

A review of robot grasping tactile perception based on deep learning

“…Zhang et al [27] introduced an innovative approach in the collection of tactile data sets by utilizing multiple sensors. This advancement enabled the acquisition of more intricate tactile data.…”

“…The contribution of Zhang et al lies in two key aspects [27]. Firstly, they optimized a ResNet10-v1 architecture that combines a convolutional neural network (CNN) [35] and a residual network (ResNet) [36] to extract features from tactile information.…”

Dynamics of the judgment of tactile stimulus intensity

“…Here, we further develop our initial inquiry as to what type of algorithm the biological nervous system might employ to perform the classification of tactile inputs, with an eye towards extending such algorithms to artificial agents (Bandyopadhyaya et al 2014, Taunyazov et al 2020, Zhang et al 2021. We examine the brain algorithms that might be at work when a trained animal, a rat in our case, classifies a tactile vibration.…”