Our sense of smell shapes how we navigate our environment, but it's a sensation that's notoriously difficult to recreate in the digital world.
To overcome this challenge, researchers have developed a device that uses an artificial intelligence e-nose to detect and analyze odors. Combined with a wearable release system called Scentronic, which uses machine learning algorithms to classify data in real time, the system passed the smell test in the lab by encouraging better interactions between humans and machines.
"This work combines innovations in hardware and software to bring new solutions to societal challenges," said Jinghua Li, senior author of the study and an assistant professor of materials science and engineering at The Ohio State University. "Our long-term goal here is to incorporate these systems into compact and personalized devices."
Used in everything from medical research to online game entertainment, human-to-machine interfaces serve as digital conduits that connect people across incredible distances. Yet the majority of these systems focus on biophysical signals, such as haptic (touch-related), audio and visual cues, rather than smell, a sense critical for perception, flavor and emotional experiences.
However, this study establishes a framework for an accurate, hands-free controlled olfactory stimulation device. After being trained on five distinct food scents - coffee, beer, red wine, orange and chocolate cake - field testing revealed that participants could identify individual odors about 73% of the time, and paired scents about 71% of the time.
The team also found that even when Scentronic was used in tandem with interfaces that could generate taste, their results demonstrated the feasibility of programmable, cost-effective odor delivery systems to create engaging multisensory experiences, said Li.
"We did see that some people are more sensitive to certain odorants than others," she said. "Despite being easy to distinguish, there was still lots of variation in intensity among individuals."
The study was recently published in Science Advances.
Despite being a prototype, Scentronic's design is so flexible that researchers created multiple versions to serve potential consumers. These include a wristband-type device that would allow users to carry it and lift it to the nose when they want to catch a whiff of a scent, and a necklace version where five odor channels align like pearls on a string, opening only after a signal is received.
Although initially fitted with the user's comfort and functionality in mind, the ease with which these sensors were adapted to further demonstrations highlights the system's seamless ability to enhance extended reality experiences beyond mere sight and sound, said Li.
Such versatility suggests Scentronic has many potential real-world applications, from supporting future assistive technologies and enhancing virtual training environments to becoming sensory aids in medical contexts, including as a way to reduce the need for frequent clinic visits, improve patient comfort and enable timely, personalized care, said Li.
"The loss of olfaction is related to many neurological diseases and infections," she said. "As a result, it's important that we create technologies to first decouple the secrets within our olfactory systems and then try to reconstruct them based on that information."
Still, while it'll be a long time before the internet is fully scratch-and-sniff capable, next steps involve advancing the work by testing other ways to incorporate the team's devices into people's daily accessories, including wearable electronics like watches or other smart jewelry.
"There are many potential opportunities for these technologies to shine," she said. "The initial demonstrations are promising, so digitized smells may arrive sooner than you think."
Other Ohio State co-authors include Chunyu Yang, Qi Wang, Antonia Akoto-Yeboah, Shaoru Cheng, Billings Amoah, Yizhen Jia, Shulin Chen, Akhil Canumalla, Kai Zhao and Christopher Simons, as well as Chen Wang from Zhejiang Laboratory in China.
This work was supported by the National Science Foundation, the Center for Brain Injury Recovery & Discovery Pilot Award at Ohio State, and the Ohio State Alumni Grant for Graduate Research and Scholarship.
