How to guarantee the cleanliness of shared cars?
Shared vehicles should experience a growth rate of 20% per year in the coming years and gradually become a new form of public transportation! However, the expectations for a shared car remain the same as in a private vehicle: the cleanliness of the interior is key. It is difficult to measure the level of cleanliness in a car after use. Evaluation is often subjective: smell, touch, etc. Vehicle cleaning is a demanding and time consuming, inefficient process. It is a challenging issue to solve: if cleanliness standards aren’t met, dissatisfied customers will abandon car-share operators.
per year of shared vehicles
in Paris by 2025
Hyundai and Total wish to develop a new detection and measurement system of cleanliness level for car-share operators. As part of the experiment, partners inquire about the most common sources of dirtiness and learn to identify those that represent a risk or an inconvenience for the passengers. The partners decided to collaborate with GreenTropism to address this issue. The startup deploys various sensors in professional vehicles to gather information on the cleanliness of the interior: images, data on stains and ambient air. The data analysis enables to detect "anomalies" such as the presence of coffee stains or grease. The aim is to understand which criteria are the most relevant in order to measure the level of cleanliness inside a car.
GreenTropism offers a solution that combines physical sensors and artificial intelligence algorithms based on spectroscopy data. The software they develop offers manufacturers simple and precise tools for the predictive analysis of organic matter or control system for industries such as agriculture, food, pharmaceuticals, cosmetics and textiles.
The sensors deployed in the vehicles detect residues of organic matter invisible to the naked eye using near-infrared light. This information is then aimed to be communicated to cleaning services. The startup first created a database from scratch by arranging stain samples in a vehicle and trained the algorithms before testing them in real conditions on UBEEQO fleets.