Machine learning-assisted flexible dual modal sensor for multi-sensing detection and target object recognition in the...

Area of Science:

• Robotics
• Materials Science
• Artificial Intelligence

Background:

• Multi-modal sensory data is crucial for robotic grasping.
• Simultaneous non-contact proximity and contact pressure sensing is challenging due to signal cross-coupling.

Purpose of the Study:

• To develop a machine learning-assisted flexible dual modal sensor (FDMS) for robotic electronic skin.
• To enable simultaneous proximity distance and contact pressure measurements for enhanced grasping perception.

Main Methods:

• Developed a multi-layer FDMS using polydimethylsiloxane, silver paste, silicone rubber, and hydrogel.
• Integrated conductive silver coils for proximity sensing (capacitance variation) and a triboelectric nanogenerator (TENG) for pressure sensing.
• Utilized the AlexNet neural network for material and hardness recognition.

Main Results:

• Achieved a 93.49% success rate for material recognition and 92.22% for hardness recognition.
• Demonstrated superior performance of AlexNet compared to other algorithms for material recognition.
• Showcased the dual perception feedback capability of the robot electronic skin with flexibility and stability.

Conclusions:

• The developed FDMS effectively performs simultaneous proximity and pressure sensing for robotic applications.
• The AlexNet-powered recognition system significantly enhances robotic grasping and human-robot interaction.
• The flexible dual modal sensor holds potential for human-robot interactions, soft robotics, and biomedical fields.