This thesis report details the work undertaken for the Mission Zero House group collective, contributing to their overarching ambition of eliminating greenhouse gas emissions from industrial processes. The primary focus of the project is the investigation of a sensor-based solution for the automated classification of battery chemical compositions – a critical prerequisite for enhancing battery recycling processes and promoting sustainable material production.

An inductive sensor was integrated into a robotic system to capture battery response data, which was analysed using a neural network classifier. Experimental evaluations indicated that, while the proposed approach achieved an overall classification accuracy of approximately 80%, there remains significant scope for further enhancement through the integration of additional sensor.

The findings of this work underscore the considerable potential of combining sensor data with machine learning methods to automate and refine the battery sorting process. This approach not only augments the precision of battery classification but also paves the way towards optimising recycling operations, thereby supporting the pursuit of zero-emission material production.

Future research efforts are recommended to further improve classification accuracy by exploring supplementary sensor modalities and optimising neural network architectures, ensuring robust and reliable industrial implementation.