Diffuse optical techniques for cocoa bean fraction segregation

Overview

This solution employs diffuse optical sensing techniques to improve the separation accuracy of cocoa nibs from shells and fines. By leveraging non-invasive and remote spectroscopic methods, it identifies distinct optical profiles specific to each component. These techniques are compatible with automated sorting systems and AI/ML applications, offering a significant advancement over traditional cocoa separation methods. The approach is designed to handle issues like caking and offers precision through volumetric sensing.

Technical specifications

Key features:

• Utilizes diffuse reflectance and fluorescence signatures to differentiate cocoa components based on intrinsic optical properties
• Employs spectroscopic profiles to identify and segregate nibs, shells, and fines
• Volumetric sensing for improved precision and sample alignment
• Compatible with machine learning and AI systems for automated sorting
• Handles caking without impacting separation accuracy

Methodology:

1. Conduct initial optical assessments using lab-based equipment to obtain diffuse reflectance and fluorescence spectra.
2. Develop metrics to identify optical signatures and optimal sensing wavelengths.
3. Simulate real-world samples and develop models for automatic classification.
4. Compare separation efficiency and yield losses with traditional methods.

Technology readiness level

Currently at Technology Readiness Level 3, this solution has been validated in controlled environments and is progressing towards the development of automated classification models. Future steps include optimizing classification accuracy and assessing the industrial feasibility of adoption.