Tactile sensor plays an important role in the human-robot interaction by providing environmental information to robots. How to effectively use the haptic information of array sensors to encode and extract haptic features for object recognition is the focus of current research. Previous studies have provided methods to construct array sensors in order to acquire tactile information efficiently. New advances in event-driven array sensors could capture spatial spiking tactile data, which include rich spatial information. In this paper, we propose a Residualized Graph Self-attention Spiking Neural Network (Tacformer) for tactile object recognition. The proposed Tacformer pays attention to both the spatial properties and the temporal sensing values of each sensor point in event-driven array sensors. Experimental results on two tactile datasets show that the accuracy, precision, recall and F1-value of the method in this paper are improved. On the EvTouch-Objects dataset, Tacformer has improved the accuracy by 8.34% compared with the GNN-SNN based a baseline approach. On the EvTouch-Containers dataset, Tacformer also has increased the accuracy by 16.17%. The Tacformer could be applied to facilitate robot tasks like grasping and sorting.
Tacformer: A Self-attention Spiking Neural Network for Tactile Object Recognition
Lect.Notes Computer
International Conference on Intelligent Robotics and Applications ; 2023 ; Hangzhou, China July 05, 2023 - July 07, 2023
16.10.2023
13 pages
Aufsatz/Kapitel (Buch)
Elektronische Ressource
Englisch
tactile perception , spiking neural network , graph neural network , self-attention mechanism Computer Science , Artificial Intelligence , Software Engineering/Programming and Operating Systems , Computer Applications , User Interfaces and Human Computer Interaction , Computer Communication Networks , Special Purpose and Application-Based Systems
Spiking Deep Convolutional Neural Networks for Energy-Efficient Object Recognition
British Library Online Contents | 2015
|Spiking Neural Network-based Flight Controller
IEEE | 2024
|Spiking Neural Network E-Nose classifier chip
IEEE | 2010
|Neural-Network Object-Recognition Program
NTRS | 1993
|