Abstract:Background Plastic Solid Waste (PSW) sorting is a procedure of paramount importance in the mechanical recycling process of plastics waste. The major limitation of the techniques relying on physical properties of plastics is the time taken for analysis and poor accuracy. Spectroscopy has been shown to be a suitable method in plastic sorting due to its atomic and molecular characterization capabilities, and ability to give results in very short time scales. However, for practical applications it is essential to translate this technique into an automatic technology, by combining it with advanced chemometric tools which can give observer independent judgement. Results The indigenously developed bi-model Laser Induced Breakdown Spectroscopy (LIBS)-Raman system with single source and single detector can record the LIBS-Raman spectral signals in single-shot mode in a total time frame of 20 ms. Out of the combinations of Principal Component Analysis (PCA) and Partial Least Squares (PLS) with Logistic Regression (LR), Linear Discriminant Analysis (LDA), Support Vector Machine (SVM), and Partial Least Squares-Discriminant Analysis (PLS-DA) based classifiers, the PLS-DA based model showed the maximum classification accuracy with 95 % based on LIBS data and 100 % based on Raman data. The reliability of the model was assessed using 4-fold cross-validation which showed a sensitivity of 90.28 % and specificity of 98.29 % for predictions based on LIBS data, and 99 % sensitivity and 99.82 % specificity for predictions relying on Raman data. Significance The results show how the combination of multimodal spectroscopy with chemometric analysis enhances the applicability of spectroscopic techniques for plastic sorting. The classification model successfully classified seven types of post-consumer plastic samples based on combined LIBS and Raman data. With the home-built software for automated prediction, the system takes less than a second to predict the plastic type illustrating the potential of the method for translation to regular routine industrial applications.

Instantaneous Material Classification Using a Polarization-Diverse RMCW LIDAR

Classification Between Digs and Dust Particles on Optical Surfaces with Acquisition and Analysis of Polarization Characteristics

Identification and Visualisation of Microplastics Via PCA to Decode Raman Spectrum Matrix Towards Imaging.

Raman Imaging Combined with an Improved PCA/algebra-based Algorithm to Capture Microplastics and Nanoplastics.

Research on Active Polarization-Based Target Detection on Sea Surface

Living in a Material World: Learning Material Properties from Full-Waveform Flash Lidar Data for Semantic Segmentation

Experiments in multiple-waveband passive polarimetric and active infrared imaging for material classification

Polarization Wavefront Lidar: Learning Large Scene Reconstruction from Polarized Wavefronts

Rapidly Measuring Scattered Polarization Parameters of the Individual Suspended Particle with Continuously Large Angular Range

Material analysis with polarization holography and machine learning

Advanced chemometric methodologies on single shot hyphenated LIBS data for rapid and reliable characterization of plastic classes

Random Neighbor Pixel-Block-Based Deep Recurrent Learning for Polarimetric SAR Image Classification

Polarization Raman microscope based on channeled spectropolarimetry

Accurate Characterization of Mixed Plastic Waste Using Machine Learning and Fast Infrared Spectroscopy

Modified Model of Polarized Bidirectional Reflectance Distribution Function Used for Light Detection and Ranging (LiDAR)

Multimodal Material Classification for Robots using Spectroscopy and High Resolution Texture Imaging

A multiparameter sensing spectroscopic system for remote plastic sorting with a single laser shot: a first information report

A Deep Learning Approach for Pixel-level Material Classification via Hyperspectral Imaging

Deep learning analysis for rapid detection and classification of household plastics based on Raman spectroscopy

All-fiber microendoscopic polarization sensing at single-photon level aided by deep-learning

Deep Learning-Based Polarization Feature Retrieval from a Single Stokes Vector