In order to build up its own data set that meets the requirements of the project, the KI Delta Learning project has started collecting data. For this purpose, a specially equipped test vehicle will collect data in Germany and abroad in the coming weeks and months.
The course for data acquisition was already set last year, while first test recordings were made this spring. After evaluating the first recordings and improving certain parameters, the regular data acquisition process has now started. Initially, trips will be made in Germany over the next few weeks, while trips abroad are planned within the next few months.
For the project as a whole, this should result in a data set that is of great importance for further work in the project. Public data sets usually do not offer the characteristics that are relevant for the research project. Thus, among other things, different deltas are taken into account during data acquisition: Examples include domain delta (different environments therefore the test drives abroad are planned), a delta in sensor technology (sensors from different manufacturers), or changes in weather conditions.
Thus, data acquisition fulfills a central role in the project: After the data is further processed and made available to all partners, all tools and methods developed in the project can be trained and evaluated once again using the new data set. This is an important step towards scalable AI for autonomous driving and therefore brings the KI Delta Learning project closer to its goal.
Image: KI Delta Learning