Scent detection and classification is an ongoing research problem that can be useful to those suffering from anosmia, particularly after the increase in the percentage of patients after COVID-19. With the ability to detect harmful paint fumes, smoke, ethanol spills, etc. in domestic as well as industrial settings, a mobile scent classification system that can map and locate the source of the odor even in low visibility situations can help prevent hazards. The issue is that current scent detection systems are immobile, extremely expensive, and inaccessible to consumers.
The ScentBot is our proposed proof-of-concept for a way of detecting and classifying a predetermined set of scents (smoke, spirits, and paint fumes) and locating the source by navigating in an indoor environment in an economical way. The robot will move in a pseudo-random manner around a mapped arena to find a scented object. If the scent is detected, the robot will then localize a path to the object and classify it as its respective scent. Our main requirement is the detection and correct classification of the scented object, along with obstacle-free navigation as shown below.