A Scenario-Based Data Catalogue for Autonomous Driving Assistance Systems Using Clustering

Autonomous driving assistance systems (ADAS) are developing quickly in terms of the safety of persons and vehicles. Test scenarios are essential for development, simulation, and validation in a typical AD sensor & safety function. As a result, the fleet car of the automobile firm collected the real-world raw data across between 1-2 million kilometers for each sensor generation. Therefore, exact data (i.e., precise data for diverse scenarios like traffic, parking lot, bridge, crossroads, etc.) is essential for carrying out development and validation. To find the precise data among millions of kilometers collected data, automated analytics solutions are enabled such as detection environmental scenes, static & dynamic road participants, scenario detection, etc. and Data catalogue will be formed for the respective scenarios based the output got from the (scenario or object detection) using clustering algorithm. The clustering is performed on the raw text data output got from the scene detection and clustered the scenarios based on the similarity. At the end data catalogue is formed for various ADAS scenarios that used to cater various development and validation requirements of higher level of autonomy.