Related Experiment Video
Updated: Feb 27, 2026
Author Spotlight: Microfluidic Channel-Based Soft Electrodes and Their Application in Capacitive Pressure Sensing
Published on: March 17, 2023
Deep Learning-Assisted Intelligent Liquid Crystal Elastomer Grippers Based on Autonomous Triboelectric Sensing
Zhengyang Chen1, Yifei Nan1,2, Lanying Zhang3
1State Key Laboratory of Optical Fiber and Cable Manufacture Technology, Department of Electrical & Electronic Engineering, Guangdong Provincial Key Laboratory of Functional Oxide Materials and Devices, Southern University of Science and Technology, Shenzhen 518055, China.
This study introduces a self-powered soft gripper using liquid crystal elastomers and dual-mode triboelectric nanogenerators for object identification. It achieves high accuracy by combining material properties and motion data, overcoming environmental interference for intelligent robotics.
Area of Science:
- Robotics and Materials Science
- Soft robotics and advanced sensor integration
Background:
- Soft grippers offer flexibility and damage-free handling but lack robust, self-powered sensing capabilities.
- Existing sensing methods for soft grippers are susceptible to environmental interference, limiting their practical applications.
Purpose of the Study:
- To develop a self-powered soft gripper with integrated sensing for autonomous target identification.
- To overcome the limitations of external sensors in soft robotic systems by creating an intrinsic sensing mechanism.
Main Methods:
- Integration of a liquid crystal elastomer (LCE) gripper with dual-mode triboelectric nanogenerators (TENGs) using fluorinated ethylene propylene (FEP) and polydimethylsiloxane (PDMS).
- Utilizing generated voltage signals (V1, V2) that encode material properties and kinematic parameters during object interaction.
- Employing a hybrid convolutional neural network-long short-term memory (CNN-LSTM) deep learning architecture for feature extraction from triboelectric/electrostatic signatures.
Main Results:
- The integrated TENG sensors successfully generated distinct voltage signals based on object properties and gripper motion.
- The CNN-LSTM model achieved 94.4% classification accuracy across 5 material categories via cross-validation.
- The system demonstrated robustness against environmental interference, a significant improvement over traditional sensing methods.
Conclusions:
- The developed LCE gripper with dual-mode TENGs provides a self-powered, intrinsically sensing solution for soft robotics.
- The fusion of triboelectric/electrostatic sensing with deep learning offers a promising approach for perceptually intelligent soft robotic systems.
- This technology has potential applications in industrial automation and human-machine interaction, enhancing robotic perception and adaptability.
