Generation of a database for AI algorithms
With the help of the creation of an AI data collection concept and test designs, surveys in real vehicles as well as continuous data quality assurance and processing, AI functions for sensors installed in the future are to be developed and requirements for series sensors for the vehicle interior are to be investigated today.
To build AI algorithms and AI-based models, labeled data from the application context is required. Especially for robust AI algorithms, Big Data is necessary, which is collected during serial operation. Therefore, sensors must be installed before AI functions can be collected from the data.
KARLI would like to investigate these AI functions for the vehicle interior as well, since this approach has not been pursued so far. Consequently, this work package will collect those data that can ensure the best possible quality of AI functions in the future.
In the Small2Big Data approach, data is collected with current series sensors and prototype sensors. Early as well as ongoing data collection in the application context serves as the basis for the AI algorithms in the KARLI project. In the work package, an AI data collection concept and test design will be developed and surveys will be carried out in real vehicles.
In addition, the exchange of this work package with WP 300, WP 400, and WP 500 helps to optimize data generation.
The goal of the work package is to collect labeled data in the Small2Big Data approach for the development of the KARLI applications. For this purpose, continuous data quality assurance, data preparation and data labeling for the use of the data by other partners will be operated.
Since the data collected in this work package offer great added value for the future, it will be made available to the scientific community in the form of a scientific publication for further research purposes.