The Olfactory Factor in AV Interiors
Aryaballe Technology’s unique sensors-on-a-chip solution aims to end the subjectivity of the human nose while neutralizing vehicle cabin odors.
Whether they’re riding in an autonomous shuttle, a transit bus, a train or a rental car, passengers often face cabin air full of “mal odors” — bad smells — including cigarette and vape smoke, pungent food, blatant lack of personal hygiene...and worse. Where the off-gassing of plastics and leather in new vehicles had been a minor issue in the past, the olfactory (science of smells) experience is increasingly a key differentiator in rider satisfaction as new mobility solutions emerge.
Ensuring a neutral-smelling passenger space is a growing focus of fleet owners and managers, particularly as autonomy becomes established in the commercial-transportation sector. Looking to a future of driverless shuttles, OEMs in the field (i.e., Cruise Automation, Waymo, Navya, ZF, Transdev, EasyMile and a host of players in China) are investigating olfaction-based sensing and notification technologies for integration into their AVs.
“Fleet owners want their vehicles to be truly autonomous in terms of the comfort of the experience,” noted Terri Jordan, executive VP of global business development at Aryballe Technologies, a Grenoble, France-based digital olfaction company founded in 2018. Aryballe’s sensor technology uses biochemical sensors, optics and machine learning to detect odor and turn it into data. “We’re taking the human being out of the cabin cleaning decision,” added Jordan, who holds a Ph.D in polymer physics and also serves as president of the company’s U.S. operation, based in Albany, NY.
Humans can sense an average of 1 trillion odors, Jordan said, but in most cases our noses cannot distinguish between them. Jordan explained that unlike color and sound, smell does not inhabit a clear spectrum, so comparing various odors is highly subjective. Aryballe’s sensor tech was originally developed for the “flavor and fragrances” industries (i.e., food/beverage and personal care/cosmetics). It evaluates the characteristics of individual scent molecules and compares them against a database of known smells using a combination of biochemistry, advanced optics and machine learning.
Its first commercial product, the NeOse Pro, is a handheld sensing tool aimed at field use. Launched at CES 2018, NeOse Pro (get it?) surprised Jordan and her team by attracting the interest of automotive OEMs and suppliers. “Their positive reaction meant that we had a new and potentially huge market,” she said.
Standards vs. subjectivity
In a 2021 SAE Technical Paper (Use of Digital Olfaction to Standardize Cabin Odor Testing in Automotive Interiors, 2021-01-0297), Aryballe scientists and engineers note that subjectivity in olfactory testing for automotive validation processes is typically a sticking point for OEMs and suppliers. For vehicle interiors, in addition to laboratory testing, human panels of “trained noses” are used to judge the smell of different components and the completed vehicle. Jordan noted that the industry lacks standardization in this area: each OEM has their own testing specifications, rating scales and methodologies.
There also are various global specs issued by SAE, ISO and VDA. The resulting overall situation regarding olfactory standards means more variation in lab testing, protracted development times and higher cost. “It’s definitely a factor that’s hindering more rapid adoption of autonomous commercial vehicles,” Jordan observed.
To help solve this logjam, Aryballe in 2019 launched the DOAC — the Digital Olfaction Automotive Consortium. DOAC members include Hyundai, Denso and Asahi Kasei, along with IFF, a stalwart in the flavors and fragrances industries. The consortium’s mission is to address industry trends related to AVs and new mobility services. Together, Aryballe and the consortium’s founding members will establish standards for measuring and categorizing odor in the auto industry and inform olfaction product development and services.
In addition to setting standards, the DOAC is collaborating to create an accessible reference database for odors. It’s also helping to evaluate prototype sensors, proof-of-concept initiatives and new versions of analytic software from Aryballe Technologies.
Aryballe also is involved with SAE’s J3263 task force that is focused on VOCs, specifically using digital olfaction methods to measure odor from materials, parts and assemblies used in automotive cabins. The task force includes plastics suppliers SABIC and Asahi Kasei as well as seat fabric suppliers.
A bio-sensing breakthrough
Aryballe’s bio-sensing technology was innovated at CEA, the French government research lab. The biosensors are peptides – short strings of amino acids, considered the “building blocks” of proteins. The human nose has roughly 400 different olfactory receptors that sense the peptides that have an affinity or reactiveness to the gases — the smells — that flow over them. Those fluidics bring the smells into our nose; the brain, in turn, interprets the pattern from these inputs and determines that the smell is strawberries, or a day-old tuna sandwich, for example.
“We graft the peptides onto a slicken surface using a chemical process. With this the sensor we’ve created recognizes a pattern based on interferometry, which is much more cost-effective and scalable than the previous SPR [surface plasmon resonance) reflective technology we were using in early development,” Jordan explained. “So, it’s not doing chemical analysis; it’s doing pattern recognition and turning the pattern into the equivalent of a picture. We then store that picture as a ‘fingerprint’ in a digital library for recall.”
Aryballe uses proprietary machine learning to “teach” the device as more and more odors are encountered and captured. Jordan said the Aryballe sensors are powerful and accurate. They’re able to distinguish between different brands of cola, between cigarette smoke and marijuana
smoke, or cigar smoke and smoldering seat fabric, for example, as well as the level of intensity of the smoke smell.
Aryballe is building the ontology, a portal, through which the vehicle cabin odor data collected by the sensor can be uploaded to the cloud, for almost real-time access by fleet operators. Jordan said vehicle OEMs will create their own parameters for odor detection.
“We can build a ‘library of smells’ for them, or maybe they use a more generic cloud-based library.” Once Aryballe and customers have validated the onboard sensor to automotive spec, “we want to be inside vehicles in three years,” she said.