SmartNanotubes Technologies

Building the AI layer for machine olfaction.

SmartNanotubes Technologies develops carbon-nanotube (CNT) electronic-nose hardware and the machine-learning models that turn raw chemiresistor-sensor response into reliable decisions.

Machine olfaction is an emerging field with very few organizations that combine a real sensing platform, deployed hardware, and honestly benchmarked models. We are one of them - and our approach is evidence-first: we publish specific models with measured, cross-device and cross-site generalization.

Our mission

Our technology is developed to digitize the world of smells and to use machine olfaction to understand the real world and to automize decision making. Our smell sensor chips are tiny, highly sensitive, low-power, and can be produced in mass. These days we are scaling up our chip production to reach hundreds of thousands annual capacity in 2027 and millions of smell sensor chips since 2028. Many real-world applications like fire detection and prevention, air purifiers and ventilation systems, food quality, elderly care, and many others are waiting for machine olfaction solutions.

Our vision

Many believe humanoid robots will change our daily lives before 2030. Robots can already see, hear and touch but they can't smell until now. This needs to change for them to safely operate in a world made for humans. With our technology we will let machines recognize smells. This page marks our first solid step toward building the Smell Foundation Model for robots. This first step is based on 5000+ of real-world smell data acquired with 100+ of our smell sensor chips in dozens of places in different countries. Starting from here, you may follow the development and growth of our Smell Foundation Model and may also join this journey.

What you'll find here

• Validated models from our machine olfactory stack, documented with full generalization rigor (leave-one-device-out and leave-one-site-out evaluation).
• Honest limitations alongside every result.
• Model weights and proprietary methods are not publicly distributed; see each model card for evaluation-access and partnership contact.

Our smell sensor technology

Each chip has 16 active independent channels. The resistance in the channels is changing in response to changes in environment (the appearance or disappearance of odors). The read-out electronics provide absolute resistance values in Ohm every second. The time evolution of 16-dimensional signal patterns follows the changes in air composition. Acquired data are labeled and used for training ML models. The ML model can be used for real-time detection of trained smells. This way our technology mimics mammal olfaction systems based on signals from multiple different smell receptors and the brain analyzing signal patterns in real time.

Find more here: smart-nanotubes.com

Work with us

Research collaboration · OEM integration · commercial licensing

Contact: info@smart-nanotubes.com