If you’re living in an urban area, chances are you’ve used some kind of car sharing service. As frequent users of shared cars ourselves, we noticed an opportunity for a simple solution to a common problem.
The Problem: Lack of Data
Currently, car sharing operators have very little information about the condition of their fleet between rides. Customers can rate the cleanliness of cars, but this means dirty ones will be an annoyance for anyone who uses them in the meantime. This results in frustrated drivers who can also have poor impressions of the car brands — not ideal when they are trying out the latest Mercedes and BMW models, for instance.
Using A.I. for Cleaner Cars
Our goal at Cleanride is cleaner car sharing thanks to A.I. Our system uses Computer Vision & Deep Learning to identify dirt, rubbish and lost belongings, and importantly without breaching customer privacy.
How Big is the Problem?
On average, each DriveNow car is driven 5 times per day, but is cleaned only once every 10 days (Source). That’s plenty of time and opportunity for junk to collect in cars.
The Cleanride team knew from our own driving experience that dirty shared cars are an issue, and a quick search of social media confirmed this. Facebook, Twitter and Instagram are full of photos of cars with cigarette butts, food wrappings and other trash left behind. And a common complaint is losing belongings in a shared car — or finding someone else’s.
We decided to find out for ourselves just how big an issue this really in our hometown of Berlin.
Time to Hit The Road!
We explored five parts of the city to check cars on different days and times. The aggregator app Urbi made it easy to find available car2go, DriveNow and driveby cars nearby.
We found that lots of cars were clean, but plenty were dirty or contained other interior issues. Based on what we saw we developed a simple rating scale:
Red: very dirty/hazard — Customers are unlikely to use or will be very unhappy. High reputational risk.
Orange: unpleasant — Customers are still likely to use the car, but their impression will not be great. Medium to low reputational risk.
Green: clean — No or negligible issues.
We found that 38% of 85 car interiors in Berlin were unpleasant, messy or very dirty.
We also recorded the location of issues in cars that we assessed as orange or red. We excluded minor things in otherwise clean cars. In total we identified 58 orange & red issues. The data shows that these are not limited to the front of the car, and that the rear seats are also a common location for issues.
While these results suggest that many people are treating shared cars with respect and are tidying up after themselves (in Berlin anyway), we also saw many examples where cars were left in a bad state.
We noted floor mats and seats covered in highly visible dog hair, which could pose a serious risk for people with allergies. We also saw issues such as cup holders filled with rubbish, bottles in doors and on seats, clothes and other belongings, sand on seats and even a bunch of sticks left in a car 🤔.
What Does This Mean?
As the car sharing market continues to grow, dirty cars and poor customer experiences could quickly become a big and costly problem for providers. With a projected 24 million active car sharing users in 2020 in Europe (source), if each customer drove one shared car a week this could mean 1.2 billion rides per year.
Research shows that attitudes to car ownership are changing and that this correlates with greater use of car sharing services (source). More and more people see cars as a service rather than a commodity to be owned. And so with shared cars becoming an increasingly common sight on city streets around the world, will people start treating them like the vandalised and mangled shared bikes that litter pavements from Berlin to Beijing, San Francisco to Sydney?
Without the ability to check cars remotely and analyse real-time data about the cleanliness of their fleets, car sharing providers will remain largely in the dark.
We think it’s possible to use Computer Vision and A.I. to identify issues in share cars, while still maintaining and respecting the privacy of customers. We think this will deliver greater efficiency for providers through automation and data-driven car cleaning. And customers will experience cleaner cars and have better impressions of car brands.
How it Works
In the first iteration of Cleanride, cameras installed in a shared car only become active once the customer has finished their ride and locked the car. After images are taken, the Cleanride A.I. identifies and flags any issues. A remote operator then reviews the images and assigns a rating. If the issue is significant enough the car is taken offline temporarily and the operator dispatches a cleaner to the car.
Over time human operators train the A.I., so that the A.I. “learns” when cars are dirty by comparing the images taken at the end of each ride. We also think that customers can be incentivised to clean up minor issues themselves, similar to how drivers are rewarded for refuelling cars or for taking them to charging stations. This would mean that the car interior is fresh and presentable for the next customer, and the driver has helped the service’s community by making the experience better for everybody.
We’ll be expanding our research in other counties, so that we have a better understanding of the problem and can compare the situation in different locations.
Get in Touch!
Want to share your stories about messy shared cars? Or do you just want to learn more about the Cleanride service? If so please get in touch with Alistair Cadman, our Chief Product Officer.