Researchers at FIT work on effective interconnection of data from smart cars and buildings

"The Cross-CPP project connects and evaluates data from cyber-physical devices, which in this case are cars and smart buildings. These devices have various sensors, readers, cameras and radars, which continuously collect large amounts of data. We try to find ways of obtaining the most useful information from these data," said Pavel Smrž, who leads the team of experts from FIT BUT. Together with his colleagues, he is in charge of processing, analysing, encrypting data and creating classification and predictive models based on such data.

Experts now use hundreds of common cars equipped with intelligent sensors and cameras to collect, design and test data infrastructure. "Volkswagen announced that it would manufacture only electric cars from 2025 on. Other applications will then be related to pairing of information on a specific electric vehicle - its current battery level and remaining mileage estimation - with smart buildings and stops that host charging stations," described Smrž. "Smart buildings will be equipped with models predicting how much solar energy will be produced in the following hours and how the buildings will consume it - for example depending on how many people there are in the building and whether they are using heating or air conditioning. The excess energy can then be actively offered to passing electric vehicles," Smrž added.

Meteorology specialists are also involved in the project. They are creating an application for hyperlocal weather forecasts using data obtained from cars and their locations. "It is possible to obtain more than just information on the outdoor temperature from the vehicles. If we know that the drivers switched on fog lights or wipers, it is very likely that there is a fog or rain at the given location. We can evaluate its intensity on the basis of the wiper speed," says Smrž, outlining the sources thanks to which experts are creating very accurate meteorological models that are currently being experimentally tested in Germany. These models will be able to predict weather with accuracy of one hundred metres. Drivers can thus get a tip on where to go in order to avoid fog.

Excess energy in smart buildings can be offered to passing electric vehicles | Author: Archive of the Cross-CP project

From the data obtained from the GPS and sensors located in the car suspension, it is also possible to create useful maps of potholes on roads. However, data protection and privacy are the main priorities of the project. Any sharing of data is protected by strict legislation concerning personal data protection. "Drivers will thus have full control over what they want to share and when they want to share it. There would probably be no one interested in stealing information on the load of car suspension, but information on the location from which people can see that I am not at home or information on speeding could be more sensitive," Smrž described some of the risks included.

If a driver of a smart car decides to share data on his/her journey and have them evaluated, he/she can obtain various benefits. In addition to accessing applications with information about weather conditions and the traffic situation and a map of charging stations, insurance companies can also participate in the Cross-CPP project. "Drivers without any accidents recorded over a longer period of time could obtain benefits in the form of insurance discounts," explains Smrž. The data collected should also be extended by information from smart stops. There are dozens of such stops in the South Moravian Region, measuring not only the ambient temperature, but also recording information on the passage of public transport vehicles.

For experts from FIT BUT, the biggest challenge of the project was to devise machine learning algorithms that can effectively process large volumes of data, thus managing situations where a large number of cars start to share their data at the same time. "We are dealing with filtering and encryption of all the data so that they meet all the requirements for anonymity and, at the same time, can be classified and processed using neural networks used by machine learning. Alternatively, we want to enable service users to share data on different bases. For example, I am willing to share my vehicle position only when there are four other cars in the close vicinity - so that it cannot be inferred that the only vehicle that could measure the temperature at a certain place was mine," described Smrž.

Although IT experts will be able to anonymise the data, it will be very important to explain in detail the possibilities of data sharing to the users of smart cars. "In Cross-CPP applications, people will always be able to choose whether they want to share their data and it will be possible to completely disable data sharing at any time. However, the ways in which people perceive sharing of their data might change over time. And so, even if we have fully automated smart cars available, some drivers might prefer manual driving so as to maintain anonymity," said the expert.

The Cross-CPP project is part of the Horizon 2020 European Research and Innovation Programme and should continue until the end of 2020. International teams of experts, including scientists from FIT BUT, now aim to test whether the proposed technologies for integration and analysis of data flows are functional. "In the eighteen months after the completion of the pilot project, we plan to work with the aforementioned companies to introduce the technologies into practice on a broader scale," concluded Smrž.