Non Human Scent

UPDATE 06/14/25: V2 PCBs Arrived!

Nonhumanscent is a platform for gas composition sensing and scent classification developed by Yoyo and I. Our goal is to develop a platform for olfactory computation, starting with an open source platform for scent classification with edge-ML models and low cost embedded hardware called Nonhumanscent V1. For now, training and programming is done through Bosch's BME-AI-Studio platform + BSEC2 firmware. We are currently working on our own scent model, training process, and open source firmware with extended capabilities and support for other sensors and processors. From there, we will build a more general olfactory computation platform.

Milestones:

• Finalize Nonhumanscent dev kit V1, test PCB, test BSEC2 integration and BLE+WiFi streaming
• Document the use of PlatformIO IDWE and Bosch-AI studio with Nonhumanscent V1
• Document the use of Nonhumanscent V1 in a variety of interesting use cases
• Begin testing additional sensors for Nonhumanscent V2
• Release a scent classification model and training process with support for additional sensors
• Develop a smell synthesis demo based on Nonhumanscent V2
• Extend the same principles of Nonhumanscent to other architectures (eg biological olfactory sensors, laser patterned high density diversified sensor arrays)

Links:

• Nonhumanscent Development Kit Repo
• Scent Context Protocol Discord
• Our talk with Ekkolapto Institute, + my slides
• Yoyo's Website