The creation of new odors by mixing existing odors has until now been a manual process based on human senses. To enable robots to do this automatically, we developed an automated odor-blending system. In this system, the target odor is first measured by the olfactory sensor MSS (Membrane-type Surface-stress Sensor). Then, Bayesian optimization determines the concentration of the liquid samples to be mixed to approximate this measurement result, and the syringe pumps are automatically controlled to actually synthesize the odor. In particular, by employing a system in which liquid samples are added to a single container and developing a suitable algorithm, the system not only reproduces the target odor, but also reduces the amount of liquid samples consumed.

Created at: 2024-06-10 03:12:03 +0900Updated at: 2024-08-12 04:30:25 +0900