/thyssenkrupp_logo_claim_d.svg){kind=logo}
carValoo – the AI for more transparency when it comes to car sharing
Bernd Overmaat
Imagine, how much easier and safer car sharing would be, if in the future you could easily check via app whether a car is accident-free or not? Hidden damage, which car dealers like to gloss over, would be visible at a glance and the consumer would be on the safe side!
With our artificial intelligence carValoo we offer a product that uses highly sensitive sensors to communicate the driving history and condition of a car in detail – for example, the residual value of a car can be determined quickly and easily.
Trust is good, transparency is better
Thanks to the detailed inspection by our AI, age and duration are no longer the only information customers can use as a guide when buying or renting a car. carValoo is immune to negotiation skills or rhetoric – because an AI deals with pure facts.
Determine residual value objectively
“Of course it is clear that a car with 100,000 kilometres on the speedometer is better than a car that has fewer kilometres behind it and the same running performance,” says Nico Schön, expert from the development team of carValoo. “But until now, buyers have not had an objective measuring instrument for this. With carValoo we’re changing that.”
Accurate value measurement instead of rough estimation
carValoo records the vehicle history. Accidents, bumps, accelerations, braking manoeuvres. All this is recorded. The name is made up of “car” and “Valoo”, derived from the word “Value”. carValoo comes in a small box and can be easily retrofitted.
With the large number of cars in a car-sharing fleet, it is especially important to be able to quickly and efficiently analyse the current state of a car.
Is the Mini before us really 5 years old? Does it really have only 50,000 kilometers down? Is it accident-free? The dealer exudes confidence, but we are not sure. Most of us aren’t experts.
“Every event has a kind of motion fingerprint,” explains Schön. “A highway trip provides different readings than a city trip, a parking maneuver has a different digital fingerprint than slamming a door, a left-hand bend provides different data than a hasty braking maneuver.”
The carValoo algorithm was trained according to the principle of machine learning and learns the movement pattern of a vehicle based on real measured values.
Quick to install
carValoo users can install the sensors independently in the vehicle. “Our solution is independent of the existing vehicle electronics and can be installed in just ten minutes,” emphasizes Schön. “If desired, we offer our customers individual installation training courses.”
carValoos easy and quick installation makes it especially user friendly.
Access to the information is also made as easy as possible for the users: The evaluation of movement profiles, damage or unusual driving and vehicle behaviour can be called up via app, so that carValoo users can check the status of the vehicles at any time. In addition, users can also use it to easily document damage events.
carValoo is compliant with data protection
We have asked how the DSGVO guidelines are considered within carValoo. Schön enlightens us: “All data collected by carValoo is encrypted and transferred to the thyssenkrupp cloud. Its servers are located in Germany and the EU. All requirements of the European data protection basic regulation (EU-DSGVO) are complied with. We only process and analyze vehicle-related data, not personal data.
Data protection is a hotly debated issue. The users of carValoo enjoy high data security because our application complies with the applicable data protection laws.
carValoo - How it all began
The carValoo project was born in 2016 as part of thyssenkrupp’s Innovation Garage. “At the beginning of the project, the question was: How can we develop new data-based business models based on vehicle data,” reports Sophie Wei, Chief Data Scientist in the carValoo team. The starting point for the considerations: the active chassis newly developed by thyssenkrupp. They record and process a large amount of vehicle data to optimally control the chassis. “We have examined other possible applications for this data,” says Wei. However, in order to be independent of the sensors already contained in the vehicle, the team decided on a separate box with its own measurement technology. This is how the carValoo idea finally came about.
Pilot phase successfully completed
At the end of 2019, carVaoo was granted its first patent. This was followed by a 12-month pilot phase with approximately 100 Car-Sharing vehicles, which was successfully completed at the end of 2019.
After a successfully completed pilot phase, carValoo has proven that it was able to identify over 90% of the existing damage in detail.
The results of this test phase were collected driving data from over one million kilometres of 20,000 car-sharing rentals. Here, carValoo was used on different types of vehicles with various types of damage. Our AI detected more than 90% of the damages that occurred.
A highlight: The successful implementation of a data pre-processing on the sensor box, which only sends relevant motion and vibration data of the vehicle into the cloud. In this way, the mobile data volume in the use case is reduced by about 90% without losing relevant data.