With the aim is to develop a flexible and scalable solution to systematically manage and analyse the huge amount of data that results from car testing, Peak Solution and PSA Groupe have agreed to collaborate on a ‘big-data’ project. The two partners will also incorporate into the project recommendations and results from the ASAM-BigODS Work Group. The volume of test data generated during the car development process has increased dramatically in the past few years.
One reason for this is the constantly increasing amount of electronic components in cars and the rise of the electric car. The data are not only provided by the cars' CAN bus systems, but also by synchronised audio and video recordings made during test runs, which can be explicitly correlated to individual driving situations during the analysis. This way, each test vehicle generates several gigabytes of data per hour for various different departments and testing areas.
Peak Solution and PSA Groupe have joined forces to lay the foundations for a more professional, efficient management and analysis of this enormous volume of data: To begin with, they intend to create a big-data-friendly data format in which measuring data can be stored along with their descriptive information. The project partners also want to define the specifications for a manufacturer-independent interface that makes it possible to use the big-data platform with existing data management and analysis solutions. The interface is to be set up in such a way that various different big-data technologies can be used for the implementation of the data access protocol.
A prototype by mid-2017
In addition to the expertise and recommendations of the ASAM-BigODS Work Group, previous expertise from the big-data IT department at PSA Groupe will also be invested into the project. The MDM-specialists from Peak Solution can draw on extensive sample data, as well as on the big-data platform BigInsights, whilst developing the system. Based on the Apache Hadoop framework, BigInsights has rolled out technologies and extended functions to process and analyse large volumes of data. An initial prototype for the solution should be available by mid-2017. Further implementation steps will then be planned before the end of 2017 based on this prototype.