Multifunctional Tactile Sensors for Object Recognition

Linglu Wang, Xiangyu Qi, Chuanbo Li, Yang Wang

Advanced Functional Materials · 2024 · 47 citations · 34 references

DOIFull text

Open access

Concepts

TL;DR

Humans identify objects by sensing shape, temperature, and material, yet replicating such finger‑like tactile sensors is difficult because compliance conflicts with multifunctionality. The study reports a robot hand sensory system capable of object recognition. The system uses a horizontal sensor that detects strain, material, and temperature via piezoresistive, triboelectric, and thermoelectric effects, and a machine‑learning classifier that achieves 98.5 % accuracy on five objects. The approach advances soft‑robotic sensory systems for object recognition.

Abstract

Abstract Humans can grab and identify objects by capturing the features of physical shape, temperature, and material. However, creating artificial sensors with finger‐like tactile capabilities is challenging due to the contradiction between structural compliance and multifunctional sensing. Here, a sensory system for a robot hand that can achieve object recognition is reported. The sensor with a simple horizontal sensing structure can respond to strain, material, and temperature stimuli using piezoresistive, triboelectric, and thermoelectric effects. A machine learning algorithm to train and classify tactile sensing signals, enabling a high accuracy of 98.5% in recognizing five objects is also developed. This work contributes to the development of sensory systems for object recognition in soft robotics.

References

34