Machine vision-enabled surface temperature mapping based on thermo-responsive cholesteric liquid crystal elastomer arrays

Abstract

Cholesteric liquid crystal elastomers (CLCEs) possess tunable optical properties and excellent thermal responsiveness, making them ideal for temperature monitoring applications. However, their broad applications are hindered due to the lack of an efficient method to correlate sample color with temperature. In this study, we developed two machine learning models: a Color-based Temperature Value Prediction Model (CTVPM) and a Color Array-based Temperature Mapping Model (CATMM), to predict temperature values and temperature maps, respectively, from images of colored CLCE samples. CLCE samples with thermal-responsive color are synthesized taking advantage of solvent evaporation-induced molecular alignment and dynamic bond-enabled crosslinking. Experimental results demonstrated that temperature values can be accurately predicted based on RGB images using CTVPM, particularly in the temperature range of 50–70 °C, corresponding to the phase transition temperature range of the CLCE samples. Furthermore, temperature distributions can be effectively mapped using CATMM. Applying machine vision to CLCE samples offers an intuitive and cost-effective method for temperature monitoring and mapping. These findings pave the way for combining CLCEs with deep learning algorithms for sensing and safety-related applications.