Non-Contact Tactile Perception in Human-Robot Interaction: Deep Learning-Enhanced Super-Resolution Spatial Sensing
Shuyao Zhou, Jikai Liang, Zhengjie Zhu, Kong Depeng, Zhiao He, Honghao Lyu, Geng Yang
AI summary
Problem
Current non-contact tactile sensors struggle to achieve high spatial resolution without complex fabrication or excessive sensor density, hindering their use in dynamic human-robot interaction.
Approach
The authors developed a flexible, layout-optimized array of 23 single-electrode triboelectric nanogenerators and processed its multi-channel signals using an adaptive spatial-temporal graph convolutional network to reconstruct precise 3D stimulus coordinates.
Key results
- 3.11 mm average spatial positioning error with only 23 physical sensors
- Accurate tracking of straight, figure-eight, and spiral trajectories
- 99.33% accuracy in non-contact gesture classification for robotic hand control
- Robust long-term stability demonstrated over 4,000 repetitive cycles
Why it matters
This approach provides a lightweight, high-resolution alternative to dense sensor arrays, advancing reliable and intention-aware human-robot interaction in unstructured environments.
Abstract
With the increasing deployment of robots in dynamic and unpredictable scenarios, it becomes necessary for robots to acquire not only contact-based but also non-contact tactile signals to enhance environmental understanding. However, current non-contact tactile sensors are largely limited to detecting or coarsely recognizing external stimuli, while achieving high spatial resolution typically entails increased sensor density and complex fabrication. This work presents a flexible sparse 2D sensor array, in conjunction with a tailored deep learning model called adaptive spatial-temporal graph convolutional network (ASTGCN), facilitating 3D spatial super- resolution (SR) perception. Built on single-electrode triboelectric nanogenerators with an optimized layout, the sensor array achieves spatial perception while providing a large perception space at low sensor density. Enhanced by the ASTGCN model, this system achieves an average spatial positioning error of 3.11 mm with a physical resolution of only 23 sensors. This research provides novel insights into non-contact haptic perception systems, enabling spatial SR tasks, including spatial trajectory tracking and non-contact gesture classification with 99.33% accuracy, where the gesture classification is used to control a dexterous hand for human-robot interaction.