Related Experiment Video
Updated: Sep 17, 2025
19:44
A Tactile Automated Passive-Finger Stimulator TAPS
Published on: June 3, 2009
13.8K
Super-resolution tactile sensor arrays with sparse units enabled by deep learning.
Depeng Kong1, Yuyao Lu1, Shuyao Zhou1
1State Key Laboratory of Fluid Power and Mechatronic Systems, School of Mechanical Engineering, Zhejiang University, Hangzhou 310030, China.
Science Advances
|July 2, 2025
Summary
Researchers developed a novel super-resolution (SR) tactile sensor for robots using sparse taxels and AI. This technology enables high-resolution touch perception, mimicking human fingertips for advanced robotic interaction.
Area of Science:
- Robotics
- Artificial Intelligence
- Sensor Technology
Background:
- High-resolution tactile perception is crucial for humanoid robots in interaction tasks.
- Current methods require dense sensor nodes, complex wiring, and processing, limiting scalability.
Purpose of the Study:
- To introduce a super-resolution (SR) tactile sensor array with sparsely distributed taxels.
- To develop a universal intelligent framework for enhancing tactile sensing capabilities.
- To achieve high-resolution tactile perception in robots efficiently.
Main Methods:
- Utilized a topological optimization strategy for sparse taxel layout design.
- Developed a deep learning model: self-attention-assisted tactile SR.
- Generated high-density pressure stimuli from a low number of physical taxels.
Main Results:
- Achieved a super-resolution scale factor exceeding 115.
- Attained an average localization error of 0.73 mm, nearing human fingertip accuracy.
- Successfully generated 2700 virtual taxels from 23 physical taxels.
Conclusions:
- The proposed framework enables facile and energy-efficient SR tactile sensing.
- This technology has the potential to equip robots with advanced embodied tactile perception.
- Surpassed current state-of-the-art solutions in tactile sensing resolution and accuracy.
