The following presentations will take place on the above date:

14:30 h: Max Frankenberg (Bachelor Thesis)
“Modeling of Human-Driven Vehicles from Real Data Using Gaussian Mixture Models”

Supervisor: Maik Pfefferkorn, M.Sc.

The presentation will be held in English.

Guests are very welcome.

Room 406A or

Zoom link: https://ovgu.zoom.us/j/67621818631

Passcode: 239117

Abstract:
The prediction of vehicle trajectories at intersections is a crucial aspect of developing successful Automated Driving Assistance Systems (ADAS) and an ongoing area of research due to the unique complexities that intersections pose for these systems. Accurate understanding of human-driven vehicle behavior is vital for the effective deployment of ADAS. This thesis investigates the application of Gaussian Mixture Models (GMMs) in the context of trajectory prediction for vehicles at intersections. Two distinct modeling approaches are compared: The first approach involves using the recent history of a vehicle to directly predict the distribution over the future trajectory, while the second approach uses recursive one-step predictions to build the future trajectory iteratively. To assess the effectiveness of the proposed models, real-world data of vehicle trajectories at intersections is used for evaluation. The findings reveal that, in general, the direct mapping approach outperforms the recursive one-step prediction approach in terms of both accuracy and computational efficiency for long-term trajectory prediction.