Automated Testing for the Release of Software Updates

Standardized IAV measurement process in production enhances reproducibility and helps ensure quality

It is necessary to perform and evaluate many measurements to introduce new engine software or a new data record or to release production statuses for passenger car gasoline engines. One of the challenges here lies in the growing complexity of interconnected systems and increasing vehicle variance. On the other hand, shorter development cycles and scarce resources are leading to a shallower depth of validation. Particularly for updates in production which must be validated in the shortest time possible, it is necessary to release software on the basis of defined standards and criteria. IAV uses a standardized procedure for carrying out and evaluating all tests and analyses. This guarantees full documentation, traceability as well as a high level of reproducibility and efficiency in measurement processes. It gives our customers transparent evidence that the tests have been conducted.

Various tools are used for carrying out, evaluating and documenting measurements automatically. Jira project management software helps IAV’s calibrators to manage and document their work in a project. For instance, they define which software needs to be tested and released in which vehicle. “It is a type of ticket system in which we document all information for the task at hand”, says Björn Becker, manager of the calibration team at IAV.

As soon as the tasks have been defined, the information is exported from Jira and imported into the INCA-FLOW tool developed by IAV. It is then clear which measurements need to be carried out and which software is affected. “INCA-FLOW holds a complete catalog of all measurements, and by configuring the tool we can also determine which measurements to perform”, Becker explains. “For example, the measurements we do for an automatic vehicle differ from those we carry out for a vehicle with manual transmission.”

INCA-FLOW automates measurements

INCA-FLOW then guides the driver through all measurements. The tool automatically takes care of starting the measurements and saving the results as soon as specific criteria are met – e.g. that a certain engine speed has been exceeded. Using acoustic and visual signals, it guides the driver through the entire test procedure and, in the background, also controls all of the parameters that are required for the particular measurement (for example, regeneration of the diesel particulate filter).

“We have already come very close to our target: In the future, just being able to operate the tool chain will suffice – the driver will then no longer need to be a calibration engineer”, Becker says. “INCA-FLOW also documents each measurement performed. We can then import the results log back into Jira.”

The team uses IAV Mara – also developed by IAV in-house – for automatically evaluating data and assessing them offline. Among other options, it can search through results on the basis of specific events and present them in graphic form. This permits comparisons between the target and actual situation and provides documentation of the processes involved. And if the client wants a log of measurement results, documentation of a release or a release recommendation, it is Jira’s turn once again. The tool can automatically generate the requested documents in the form of PDF or Word files.

Knowledge integrated from all of IAV’s divisions

Automated testing permits full documentation and traceability as well as a high level of reproducibility of measurements performed. “This approach also enhances quality because it is based on defined release tests and criteria”, says Becker, summing up. “The knowledge of all IAV divisions is channeled into the process. As a result, the driver doesn’t need to have any specific knowledge.” In future, it is planned to add the use of touch screens and automated evaluation of measurement results. If a limit value is violated during a measurement – e.g. by excessive temperatures – the system automatically interprets the measurement as being invalid and displays a graph of the measurement range concerned.