Initial situation
In endurance testing of new vehicles, data input of the driver's subjective perception is carried out during or after a test drive. This process is carried out using manual questionnaires, Excel tables or other manual/semi-automated methods and is time-consuming, costly and error-prone. Additionally available data such as environmental, driver and vehicle data are not fully included in the evaluation of abnormalities. In addition, the tests are not linked across fleets and vehicle platforms in order to gain further insights from endurance testing.
Target
Therefore the Chair of Automotive Engineering of the TUM wants to develop a cloud based software tool in cooperation with two industrial partners within the research project Firefly. The goal is a highly automated testing process with a Digital Test Assistant. The test planning and evaluation shall be simplified and the comparability across fleets and platforms shall be ensured. The recording and analysis of additional data (environment, driver, vehicle) should increase the significance and quality of the test run. The Digital Test Assistant uses statistical evaluations and methods of Machine Learning (ML) to analyse measurement runs of the vehicle test in real time, combine them with data from other test runs and dynamically adapt the further test run.