Sound and image are no longer challenges. Some robots, especially in the medical field, had managed to transmit a kind of touch, something that enabled them to perform surgical interventions. But the smell was something he resisted. Until now.
“We have already achieved Scent Teleportation in our laboratories and document this process in a video from Scent Teleportation – the developers point out -. Our first successful attempt involved a simple slice of coconut. One could describe what happened that day as we send the aroma of that coconut from one side of the laboratory to the other. But that description elides a much more complex underlying process of digitizing and reconstructing the molecules that make things smell the way they do.”
Osmo scientists selected a scent to teleport and fed it into a machine called GCMS (gas chromatography-mass spectrometry). If it is a liquid, it is injected directly; If it is a physical sample, such as a plum, a variant of GCMS is used, in which the aroma is trapped in the air around the object and absorbed through a tube. The GCMS identifies the raw data, which can be interpreted as molecules, and uploads it to a cloud. Over there become a coordinate on the Main Smell Map, an AI-powered tool that can predict what a particular combination of molecules smells like. This formula is sent to a robot that treats it like a recipe and mixes different aromas to replicate the sample.
The last step is to compare the original sample and the “copy” to ensure that the recreated scents align with their original scent families. The greatest difficulty faced by those responsible is that some molecules are so subtle that they barely register on sensorsbut they still significantly affect the overall aroma. Sulfurs in tropical fruits are notoriously difficult to detect and reproduce. Some molecules involved in everyday aromas remain unidentified.
“Our ongoing data collection process is an effort to reduce this number of mystery molecules and find new ways to recreate them –-. Every time we solve one of these difficult cases, we feel like we are making important progress. We’ve already amassed the world’s largest AI-enabled scent databank. These Data is essential to train our algorithms; refine their noses, so to speak.”
But turning all that data into usable formulas still takes some human guidance. Simple algorithms that use existing data are used to decode aromas; machine learning deformulation models to predict formulas from GCMS data and the Odor Map.
“We subject these tools to constant stress tests in monthly aroma teleportation tests that help us refine our processes – Osmo concludes -. Soon we will begin holding demonstrations open to the public. We will ask them to choose a flower or fruit from a large bouquet and then allow them to reflect on its smell, while we analyze a sample in our laboratory. In a single session, we will recreate your special aroma and diffuse it back to you. And then, the moment of truth: knowing their opinion.”