Hardness-and-Type Recognition of Different Objects Based on a Novel Porous Graphene Flexible Tactile Sensor Array

Area of Science:

• Robotics and Materials Science
• Human-Machine Interaction
• Sensor Technology

Background:

• Accurate tactile sensing is crucial for advanced human-machine interaction.
• Existing tactile sensors often face limitations in performance and material properties.
• Developing novel sensor materials and recognition algorithms is essential for progress.

Purpose of the Study:

• To design and fabricate a novel porous graphene flexible tactile sensor array.
• To develop and evaluate a Residual Network (ResNet) model for object recognition using tactile data.
• To compare the performance of ResNet against other machine learning models for tactile sensing tasks.

Main Methods:

• Fabrication of a porous graphene flexible tactile sensor array.
• Collection of tactile time sequence signals from objects under slight compression (2 mm).
• Implementation and training of a Residual Network (ResNet) model for classification.
• Comparative analysis with Multilayer Perceptron (MLP), LeNet, Multi-Channel Deep Convolutional Neural Network (MCDCNN), and ENCODER models.

Main Results:

• The porous graphene sensor array achieved high performance in detecting tactile features.
• The ResNet model attained average accuracies of 100% for hardness recognition and 99.7% for object type recognition.
• ResNet outperformed MLP, LeNet, MCDCNN, and ENCODER models in both hardness and object type classification tasks.

Conclusions:

• The novel porous graphene tactile sensor array exhibits excellent perceptual capabilities.
• The ResNet model is highly effective and precise for object hardness and type recognition using tactile sensor data.
• This combination significantly advances tactile sensing for human-machine interaction applications.