Patent Filed: Olfactory Sensing Technology

Recently, we filed a patent covering a core part of our artificial olfaction platform. The invention focuses on an olfactory sensor that detects chemical smell fingerprints and converts them into digital signals that can be analysed by machine learning models.

The patent builds on our early work investigating molecularly imprinted polymers (MIPs) for sensing gaseous volatile organic compounds (VOCs). Over the past year, our research has centred on understanding how polymer design influences sensor selectivity - one of the fundamental requirements for building practical electronic noses.

As part of this work, we have generated validation data from more than 170 MIP formulations with different polymer backbones and chemistries. This dataset has enabled us to systematically evaluate how changes in material composition affect sensing behaviour and to improve the selectivity of our sensor array. Biological olfactory systems achieve discrimination through many receptors with different response profiles, and our approach follows a similar principle by engineering complementary sensing materials rather than relying on a single highly specific sensor.

The work has also highlighted the importance of data. As our sensing platform has matured, we have consistently observed that improvements in sensor quality translate directly into better-performing AI models. High-quality, well-labelled olfactory data remains one of the main bottlenecks in developing robust machine learning systems for smell, making sensor development and data generation closely linked challenges.

This patent represents one component of the broader platform we are building to digitise smell. By combining advances in polymer chemistry, sensor hardware and machine learning, we're creating sensing systems that extract meaningful information from complex chemical environments.

We will be sharing further technical results from the computational and ML aspects of this work over the coming weeks.