PILOT : SUSTAINABLE CONNECTED CARS IN FRANCE

The pilot analyses data sources such as vehicle specifications and real-time information of different parameters and GPS locations, which are transmitted through telematic dongles within the cars to a networking cloud (e.g. speeds, start-up time, final time, fuel consumption, trouble code frames and GPS position). From this cloud, these features are sent to a Big Data infrastructure for their analysis. Additionally, information on roads and weather conditions are available.

There are many factors that contribute to fuel consumption like trip statistics, type of road, intensity of traffic, weather conditions, vehicle or driving performance. These factors are being measured or quantified. A set of advices or suggestions to drivers has been defined to improve situations where the fuel consumption indicator can be improved.

When an issue is detected, e.g. a car spends a lot of fuel, the driver behaviour statistics are poor, or a breakdown is detected, a finer-grained analysis can be obtained displaying the trips dashboard by the fleet manager.

After a trip is finished, its main statistics such as duration, fuel consumption or distance are received by the system. These statistics are the building blocks to ultimately calculate the performance indicators. Tips to avoid poor performances will be issued through a notification system, so drivers can improve their driving by attending these advices.

We have developed a way to detect anomalies that involves the comparison of data from different vehicles and dates, using a key dataset containing the vehicle specifications, such as model, dimensions and fuel consumption. By grouping vehicles with a common specification, e.g. same model, statistics or measurements for groups can be obtained. Anomalies could be detected by comparison of the statistics or indicators of a vehicle with the indicators of the group of vehicles it belongs to. A candidate anomaly can be obtained comparing figures for a same day. But an abnormal figure can be circumstantial, e.g. weight, weather conditions, roads could be different.

The system can extract the codes relative to malfunctions, warnings or alerts. A breakdowns service will give, when it is invoked, the severity and description of every code included in the request. Maintenance information (e.g. mileage and next revision date) is also considered in the severity evaluation. Performance comparations with group of cars (e.g. cars belonging to a same model) or with historical data is used to detect maintenance needs.

The following main objectives were addressed by the Sustainable Connected Cars pilot:

• Deployment of a Big Data infrastructure with descriptive and predictive analytics capabilities.

A machine, configured with Big Data analytics capabilities, has been deployed.

• Development of API services to carry out data injection to the Big Data infrastructure.

The following APIs have been implemented:

• Car sensors API: Data from cars received in the cloud is pushed to an access point of the Big Data platform. Additionally, static information about cars such as brand, model or type of fuel can also be obtained from the cloud.
• Car geo-tracking API is also operative.
• An API to indicate the severity of malfunction messages from cars.
• Development of a generic visualization tool for data sets analysis.

The objective for this tool is to examine and compare various car magnitudes such as fuel consumption, distance, eco-driving parameters revealing the behaviours of the drivers, error codes generated or average rpms. A dashboard to analyse trips, included the above-mentioned parameters, and the notifications issued has been developed.

• Breakdowns estimator system.

The system can extract the codes relative to malfunctions, warnings or alerts. A service by Autoaid will give, when it is invoked, the severity and description of every code included in the request. Maintenance information (e.g. mileage and next revision date) is also considered in the severity evaluation. Performance comparations with group of cars (e.g. cars belonging to a same model) or with historical data is used to detect maintenance needs.

• Emissions reduction system.

Although the main goal of this system is to reduce fuel spends, and hence emissions, we preferred to name it as “Emissions reduction system”, better than “Fuel control system”, or similar names, because there are some transversal KPIs in the TransformingTransport project belonging to the environmental quality category. Also, the breakdown system cannot be discarded to reduce emissions. There are many factors that contribute to fuel consumption like trip statistics, type of road, intensity of traffic, weather conditions, vehicle or driving performance. These factors are being measured or quantified. A set of advices or suggestions to drivers has been defined to improve situations where the fuel consumption indicator can be improved.

The expected results derived from the achievement of the objectives are as follows:

• Reduction of fuel consumption: Due to a better maintenance and improved driver behaviour.
• Reduction of emissions: Derived from the less fuel consumption.
• Safer driving: A better driving implies a safer driving. Additionally, better maintenance or vehicle condition contributes to safety.

Better maintenance: Through the earlier detection of problems with the car.

The main KPIs of the pilot are:

• Reduction of GHG emissions: The amount of emissions of CO₂ from a set of cars is a key indicator of the pilot, as road transport is the second biggest source of greenhouse gas emissions in the EU. The pilot will try to show that drivers observing the advices about driving behaviour that they receive after a journey will consume less, and therefore, will emit less CO₂ emissions to the environment. Another factor that contributes to the reduction of emissions is the state of the vehicle. For instance, vehicles with a deficient tire pressure are prone to consume more fuel. We give indicators to the fleet managers about the different messages related with technical issues detected by each car, with their severity and an estimation of the most problematic cars: Quantify % reduction of grams of CO₂ per km and vehicle.
• Reduction of NOx emissions: NOx emissions are another important indicator to be considered due to links to human health problems: Quantify % reduction of grams of NOx per km and vehicle.
• Reduction of number of messages related with breakdowns: One key factor related to safety and maintenance is the probability of register messages that are indicators of problems in a car: Quantify % reduction in the number of messages per vehicle and 10000 km.
• Reduction of the number of harsh accelerations (decelerations): A higher number of harsh accelerations and decelerations is a key factor that contributes not only to more environmental emissions, but to a less safer driving and poorer car maintenance. These numbers are registered for each trip and, if necessary, communicated to drivers through the notifications system: Quantify % reduction of the number of harsh accelerations-decelerations per km and vehicle.